I would like to spread awareness of the Condorcet method. To voters, it appears the same as instant runoff voting, but Condorcet is much harder to vote "tactically" under (and there's less need to, too). The biggest problem with IRV is that "in some situations, if a voter or group of voters decides to rank a preferred candidate lower, it can result in that candidate winning the election, whereas if they had ranked the candidate higher, according to their sincere preference, that candidate would not have won."

There are tradeoffs in every election system, but I believe the tradeoffs are lower using the Condorcet method. Determining a winner is more complicated than in IRV, but it's still an algorithm that can be coded into law fairly easily.

Thus ends my attempt at raising awareness. Thanks for reading!

Yes, there are a number of other voting techniques, including approval voting and the Borda count. Arrow's Theorem tells us that none of them are perfect -- it's possible, for any of them, to come up with weird situations where something undesirable happens.

I don't want to get into an argument over whether IRV is the best choice, but I think we can all agree that plurality voting is pretty much the worst choice. Since there's an actual bill in the legislature that would enable (not mandate!) IRV, I'm on the bandwagon.

Curiously, Anne's comment above is exactly the response I got from the Secretary of State's office. My reply is below

Regarding voters changing their minds:

There's always the possibility of new information coming along. Any voting system asks voters to commit on election day, even though some scandal might come out in the next day's paper.

Are you proposing that the voters might want to take into account how others voted in previous rounds? I believe this would be a bad thing, as it leads to insincere "strategic" voting.

For example, suppose there are four candidates, A, B, C, and D, and I prefer them in that order. On a first round of regular voting, the results are:

A 36% B 25% C 25% D 14%

There is no majority, so D is eliminated and there is a second round of voting among the remaining candidates. Even though A is still my favorite, I would be wise to vote for B to make sure that C is eliminated. This is a dishonest vote. Instant Runoff Voting effectively eliminates this sort of abuse.

Regarding "voting twice":

This is a misinterpretation. Every vote is counted exactly once in every round. Suppose, for example, that you vote for a major party candidate and I vote for a third party candidate. In the first round of IRV, my candidate is eliminated. My one ballot is moved to one of the other piles. In the second round, my vote is counted for that candidate AND your vote is counted again. One person, one ballot, one vote per round.

One might more reasonably argue that the standard plurality voting scheme violates the principle of "majority rules" because a candidate can win with a plurality even if a majority of the voters think that candidate is the worst one.

Regarding undervotes:

These are non-issues. If a ballot doesn't have a 2nd choice listed, go on to 3rd, 4th, and so on, until a vote is found. If no votes are found, discard the ballot; this is just like an "undervote" in traditional voting systems.

Unfortunately, Condorcet voting has its flaws as well. Number one, for someone like myself who didn't want to support John Kerry because he voted for the invasion of Iraq, voted for the PATRIOT Act, and voted for the North American Free Trade Agreement, I would not want my vote to count to help elect John Kerry unless all of my other choices, Nader, Cobb, and Badnarik, had been eliminated. Condorcet helps to elect your second and third choice vote, whereas Instant Runoff Voting (IRV) only allows support for your highest-ranked candidate who has a chance to win. Borda does this as well, which is why i don't like Borda, personally. I want my vote to count in rounds amongst my preferences, not for all the candidates i ranked.

With IRV, your vote only counts towards candidates who can possibly attain a majority. Condorcet is superior in its mathematical application, but it is terribly confusing and very difficult to explain in 5 minutes to someone highly educated.

Anne posted that counting for different people at different times violates the Supreme Court of the United States' standard of "one person, one vote." This is NOT accurate at all, as for one, your vote is ONLY counting for one candidate at one time, not to mention her interpretation of that standard. The standard was actually meant to state that each person must have an equal vote and equal power. IRV held up under court challenge in Michigan, the case and judgement are available here: http://www.fairvote.org/library/statutes/legal/irvlegal.rtf . Since there is case law clearly conflicting with Anne's statements, I think we can determine that point is not valid.

Second point Anne raises is that voters who rank candidates who end up with more votes don't get their vote counted for second choices because their candidate is not eliminated. That is the point: that is their highest preference, so i am not clear why they would want their second choice counted. All who have theirs counted did not want the candidate it would be counted for more as they would have ranked their candidate higher. This seems a pretty invalid argument against IRV, to be honest.

The third point Anne raises is that ballots that have their candidates eliminated and the votes are exhausted cause a problem in determining a majority. This is also not accurate as they are considered "exhausted" and removed from the total. A majority is still determined as they don't count anymore in the further rounds of counting since the voter has indicated no preference of the remainders, or rather has indicated to prefer that none of the remainders are elected, in actuality, by not voting for any of them. The voter has every right to do so.

Finally, I agree all methods, including approval voting, my LEAST FAVORED of all the alternatives, are preferable to pluralities. I like IRV and Condorcet the best, though i really think Condorcet is too complex to every implement. And Borda presents situations where someone that doesn't have a majority or even much of first choice votes could win. Electing someone with less than say 10% of the first choice votes would cause a horrible outcry by the losers that I don't think Borda would survive. Borda would be blamed for the loser losing and we would lose our precious reform.

Here are my concerns about instant runoff voting 1) The 2002 Gov. campaign was mentioned. What if someone thought of Tom Cox and Kevin Mannix as anti-tax blowhards who hadn't proposed any specific solutions? This idea implies that Cox, Mannix, Kulongoski (or Nader, Gore, Bush in 2000) were of equal quality and that one or more candidates hadn't said something that offended any voters. What if they weren't equal in the eyes of voters?

2) Any political argument quoting a theorem bothers me. Has the person quoting the theorem gone door to door for a candidate? Can this person imagine going door to door explaining this system, esp. to people so busy with work and family that if you came to their door they'd say "you've got 90 seconds--what do you want?".

3) I was once involved in an organizational election with this kind of a system. I was going to vote for the 3rd candidate as my first choice, until I discovered this person was making a side deal with one of the more well known candidates. I suspect there are lots of ways to "game" the system.

I don't see how this helps.

The One True b!X points out that instant runoff voting (IRV) is allowed by Article II, Section 16 of the Oregon constitution. Unfortunately, it’s possible to interpret that section and various laws in such a way as to prohibit local jurisdictions from using IRV. I’m not saying that’s the right interpretation, just that it’s an interpretation some people make. This leads to election officials making black-and-white pronouncements like the following:

“Oregon law provides that ‘the person receiving the highest number of votes shall be nominated or elected.’ (ORS 254.065) Statutes would have to be amended to provide for Instant Runoff Elections in Oregon.”

 - John Kaufmann
   Director of Elections
   Multnomah County

The above quote is from written testimony Kaufmann submitted to the Multnomah County Charter Review Committee on 4/13/04 when it was considering whether IRV should be used for county elections. It should come as no surprise that the committee dropped the subject: The director of elections told them they couldn’t do anything, and they believed him. It doesn’t matter whether he was right or not.

Note especially that he didn’t say the matter was inconclusive or subject to interpretation. He presented one interpretation, and he presented it as absolute fact.

Something along the same lines happened when Eugene was considering IRV in 2001. The prospect of an expensive court battle with the state elections division siphoned important energy from the effort to educate people about IRV.

The bottom line is that while it might be possible to implement IRV in Oregon without first changing state law, the process of doing so would be (already has been) like banging our heads against a brick wall. If we can pass a state law that explicitly allows local jurisdictions to use IRV, that will help everyone focus on the real question: whether IRV is better than what we’re using now.

Because it always elects the candidate exactly in the middle of all voting, Preferential (ranked) Ballot [PB] is top-dead-center-counter-extremist & thus more anti-terrorist than all the many recent retrenchments combined. While Preferentiality would be equally useful for all else its real power is most clearly shown in the case of the impending civil war in Iraq. Unless the Iraqi Parliament comes to select its PM by PB, it may not hold, & the world will be in danger of going to war over some oil well. Because it gives the Iraqi minorities a real say in which Shia gets chosen, PB is the only thing that will encourage them to support any plan more than perfectly inadequate confederation. Because it gives all combinations of programs, & not just parties, an equal chance of winning it is the only thing that is really just, for Iraqis & everyone else. Because it will always provide a market-based alternative to all proposals for justice offered by the left & vice versa, it is something that both sides could & should support. Be a light to the world & help put this powerful idea in time to as many as possible. We imagine running on the SINGLE ISSUE of Preferential Ballot & allowing a citizens advisory board based on “Organized Communications” to guide our vote for the rest. Everyone should be sure to do the same, from the most local on up. $15,000.00 will get a full page add in USA Today, which could be enough to put it to everyone. Organized Communications is small randomly assigned discussion groups electing reps to higher & higher levels by means of Preferential Ballot until one small group at the top necessarily names the next winner of the next election, thereby installing preferentiality virtually instantly, & inevitably globally, no matter what ladders are pulled up next.
The “additive” (as opposed to the more skewed, & more widely espoused, “eliminative”) form of Preferential Ballot is the ranking of all candidates by each voter. The first choices are counted & if noone has 50% then the next choices are added in, & so on, until someone finally does. It is the sole unchangeable plank & bylaw of the Preferential Ballot Party, the only practicable third party, whose self-organizing principle is Organized Communications. Please see www.preferentialballotparty.org. How can we ask it of others if we do not have it ourselves?

Zoe Zidbeck, POB 38245, Albany, NY 12203, [email protected], www.preferentialballotparty.org

Replying to Anne:

I believe IRV is good for major parties, minor parties, and the voters. As I said in my original piece, we've seen two major elections lately where the candidate that was least liked by a majority of the voters won. That's not good democracy.

The argument from Dana Dickson (the Utah link above) that "will announce who he supports to his supporters and they will vote accordingly for that candidate to be the #2 choice on their IRV ballots" is a gross overestimate of how organized and disciplined third parties are.

This site, which likes Condorcet voting, argues that IRV is actually BAD for third parties:

http://electionmethods.org/IRVproblems.htm

(I'm not sure I agree with that site, but there it is.)

I hate to get into an "is too, is not" debate over one person, one vote, but the statement that the major party votes only count once is simply false. Please have a look at this simple example:

http://www.pacificgreens.org/issues/irv.shtml

In the runoff, the votes for the minor candidate (Bob) are indeed counted again. However, the votes for Akiko and Carlos are also counted again. Every vote is counted the same number of times.

As a final comment on "gaming the system," one must imagine that nobody would know how to do this better than political scientists. The American Political Science Association uses IRV.

(The 32,000-member Mathematical Association of America and several similar organizations use approval voting, but that's not on the table in Oregon this year.)

Incidentally, what would really give third parties power is proportional representation: if your party gets 5% of the vote, you get 5% of the seats in the legislature, even if you didn't win in any particular district. Again, no bill this year, so it's not an issue.

Anne: Yes, of course Kaufmann’s quote was correct. That wasn’t my point. My point was that he presented as fact one interpretation of what his office could legally do, when there is at least one other well-known and very defensible interpretation.

The Oregon constitution says IRV may be enacted “by law”, and there is judicial precedent that local ordinances are “laws”. Therefore, the constitution can be interpreted as giving local jurisdictions the right to enact IRV should they choose to do so. Under such an interpretation, it really doesn’t matter whether ORS 254.065 allows IRV or not, because state laws are subordinate to the state constitution.

So your claim that “IRV would require statutory authority and some statutory changes to implement” is by no means a foregone conclusion.

It would still be good to pass a state law that explicitly allows local jurisdictions to use IRV, because that would clarify the situation.

Can't we just give each voter 10 colored dots for them to distribute as they like next to candidate names on butcher paper that we hang up around town?

Lest we forget, not every election has more than 2 candidates. People who vote 3rd party sometimes control the outcome (Smith over Bruggere, 1996, Riley vs. Gallegos and Burley vs. Stiegler in 2004, for example) but in many cases there are only 2 people running. There is a 33 majority in the House. 7 of those races were decided by less than 1000 votes. I don't see how IRV would change that and maybe more energy should be given to winning the narrow 2 candidate races.

BTW, did you know that freshman Betty Komp won by a larger margin than Speaker Minnis? Both of those were 2 candidate elections.

Election returns are fact (unless of course someone tampered with the vote count). IRV remains a theory, it seems to me. And why would ballot security be stronger in IRV than currently in Oregon?

As a third party member, I don't see the spoiler possibility as a useful tool. While it may be possible to play kingmaker in a well-disciplined congress or parliament, nobody is in a position to offer the votes of an American political party in return for some policy promise. Many third parties wouldn't trust a major party to keep such a promise.

More to the point, we don't like being thought of as spoilers. We have issues we'd like to talk about, but the media won't talk about anything but spoiling. In 2000, Nader couldn't get a word in edgewise without a TV host asking him, "But aren't you worried about taking votes away from Gore?"

Now, for added information, here's the site plugging approval voting:

http://www.approvalvoting.org/

Interestingly, Arrow's Theorem doesn't apply to approval voting, because it's not a ranked-ballot system.

Coming late to this discussion, I will say only that I find the all the objections to IRV unconvincing. I am a Democrat.

Gavin: you may be right. On which of Arrow's criteria does approval voting fail?

Here's a nice page that includes (about a third of the way down) the Big Table we've probably all been looking for:

http://lorrie.cranor.org/pubs/diss/node4.html

Gavin:

Thanks for the example.

First off, I of course don't mean to say that approval voting disproves Arrow's Theorem. A theorem is a theorem, after all -- once it's proven, there's no going back. However, if the theorem is of the form, "If a voting system is a ranked voting system, then it can't meet all of the following criteria," then a non-ranked voting system might still meet all of the criteria. Approval voting proponents seem to claim that this renders the theorem irrelevant to approval voting. I should look into the precise definition of the theorem more closely (and consult a colleague who has more expertise on this stuff) before making this claim.

Moving on to your example, I don't believe this is a violation of IIA. The outcome hinges on whether the last voter approves K. K is one of the contenders, so not irrelevant. If the outcome between K and B depended on whether someone approved N, that would be a violation of this criterion.

Your example does raise a legitimate complaint, though: under approval voting, voters must decide where to draw the cutoff line among their preferences, and it may matter. I believe approval voting is strategy-proof provided that everyone's preferences are dichotomous, i.e., each voter thinks, "I find this set of candidates acceptable, this set unacceptable, and beyond that I don't care."

Returning to the origin of the thread, I'll reiterate: almost anything is better than plurality voting, and IRV is the reform we have a shot at this year.

Gavin:

Thanks for the example.

First off, I of course don't mean to say that approval voting disproves Arrow's Theorem. A theorem is a theorem, after all -- once it's proven, there's no going back. However, if the theorem is of the form, "If a voting system is a ranked voting system, then it can't meet all of the following criteria," then a non-ranked voting system might still meet all of the criteria. Approval voting proponents seem to claim that this renders the theorem irrelevant to approval voting. I should look into the precise definition of the theorem more closely (and consult a colleague who has more expertise on this stuff) before making this claim.

Moving on to your example, I don't believe this is a violation of IIA. The outcome hinges on whether the last voter approves K. K is one of the contenders, so not irrelevant. If the outcome between K and B depended on whether someone approved N, that would be a violation of this criterion.

Your example does raise a legitimate complaint, though: under approval voting, voters must decide where to draw the cutoff line among their preferences, and it may matter. I believe approval voting is strategy-proof provided that everyone's preferences are dichotomous, i.e., each voter thinks, "I find this set of candidates acceptable, this set unacceptable, and beyond that I don't care."

Returning to the origin of the thread, I'll reiterate: almost anything is better than plurality voting, and IRV is the reform we have a shot at this year.

It's fun seeing so many Reedies contributing to this thread. (First Gavin, then Paul, and now myself) I have to say, I find Anne's comments about one person one vote rather surprising. The truth about any social choice (and that’s really what all voting is) is that it is a contest between n+1 options for n winners. When I am choosing who I want for president, I know that in the end there will be someone who comes in first, second, third, and so on. Under a plurality voting system, I either make the choice to vote for one of the two individuals I predict will come in first or second, or I make the choice to vote for someone I do not expect to have a chance at winning. In a race for one seat, the only choice that matters for any given voter is the choice between the first place and second place candidate.

The fundamental concept which I think Anne misses is that any votes for candidates other than the winners do not matter. In a gubernatorial election, unlike in horseshoes, close doesn’t count. Thus, under the current system, we force voters to make a choice between voting for the person they prefer, and voting strategically in order to ensure their most favored outcome. Forcing voters to misrepresent their preferences in order to get the outcome they desire is bad for democracy, because it results in a misrepresentation of what voters actually desire.

What it demonstrates is that the election system used is not accurately allowing voters to both have their voice matter, and honestly express what they think. Anne believes like if you want your voice to matter, you should have to pick one of the two major parties, because such system allows the two major parties to retain control over the elections. I believe that the two major parties should have control if and only if the voters actually agree most with what they say. A majority of likely voters right now would vote for the democrat in a congressional election (according to the last issue of polling report I looked at). Yet, we have a Republican congress. This is due largely to the fact that we have districts, but also partly due to the fact that some voters cannot express their true preferences on the ballot and at the same time have a say in the outcome.

I think the reason Democrats fear IRV is simply numbers. Anne obviously has ties to the Secretary of State’s office, who I’ve found out from others has many high ranking officials opposed to IRV (though I’m not sure about preference voting generally). The fact is, in Oregon, the Libertarians outnumber the Pacific Greens, and because, if you look at the last State House and Senate elections, there were many more Libertarian candidates than Pacific Green candidates, the Democrats were able to garner more third party support from green voters than the republicans were from libertarian voters. It is partisan politics at its worst, however, to oppose a better election method simply because it helps you get elected. I’m a Democrat, and I want Democrats to be in power because they’re making schools better and helping working people, and taking care of the environment a lot better than the republicans are.

But beyond being a Democrat, I am a democrat (small d), and that means I support democracy, for better or worse. Just like we should build the best possible schools so everyone can get a college education and everyone’s lives can be better, we should have a voting system where everyone can be honest about their preferences and everyone can have their opinion matter. Plurality voting does not do that. Probably one of the worst arguments I have heard attributed to members of the secretary of state’s office is that we should oppose IRV because voters are too stupid to rank their choices. I’ve run the elections at Reed College for two semesters so far, (I’m a sophomore) and I just changed our system of electing our Senate from a plurality system to a preferential system (STV). Instead of Reed students having to vote for five candidates because there are five seats open, students now get to rank their preferences, and can give multiple individuals the same ranking.

While I’m sure some people will screw this up, people screwed up when they had to check off five people, instead checking off four or six (and in the latter case having their ballot thrown out). And while I’m sure some people wont understand the method by which the votes are counted, the important thing is that they are counted, and are counted fairly. The nice thing about STV is that it that it is particularly resistant to strategic voting, and so for the first time, Reed elections will actually be an expression of what the voters actually think, rather than what they are forced to mark in order to get their most favored outcome.

I think this serves as a good analogy to Oregon. IRV is not the perfect system (Condorcet is, but it doesn’t always produce a winner, however, there are modifications of Condorcet like Ranked Pairs which fix this), but IRV is at least fairly explainable to people, and it suffers from the same failings that an actual runoff election would have. (In a plurality runoff election, I can vote for the person who I think will be a weaker challenger to my favorite candidate (say, voting in the Republican Primary for Mannix in 2002, rather than Kulongoski in the Democratic primary) because I think Mannix will have a harder time beating Kulongoski.) In both IRV and the runoff example I just described, I am taking a risk by not voting for or ranking highest my most favored candidate, and it is a risk I should not have to take. IRV does not give voters the ability to vote twice anymore than a runoff or primary system gives third party voters the ability to vote twice, once for the third party candidate in the primary, and once for the republican or democratic candidate in the general.

In the last Reed election, we had 7 candidates for Student Body President. Around 615 votes were cast, and after counting first preferences, the first place candidate only had 33% of the vote. We could have just left it there, called him the winner by plurality, and called it a night. But instead, all the voters who didn’t vote for him or the person who came in second got the opportunity to help make a choice which mattered, just like the people who voted for the first and second place candidates did. And this is what Anne misses. We aren’t giving third party voters two votes, we’re just giving them one vote. A vote which up till this point, they haven’t had. In the end, the person who got the most first preferences ended up winning the election, but only by 15 votes. And because we had IRV, a majority of people ended up actually deciding the election, rather than just 33%.

If our traditional 2-party system and plurality voting worked, why do we have over 30% independent/third party registrants and over 40% (most of the time) not voting?? Let's face it...our democracy is broken and needs to be fixed. We could start by junking party primaries that disenfranchise the +30% (and over 50% of people under age 35, as Phil Keisling likes to point out) but IRV is even better and should be encouraged. We should have PR, but America ain't a parliamentary system, but while we're at election reform, anybody for a unicameral legislature (one house) like Nebraska's?

Oh, and Kari...I recall that dot system at Hood River didn't work so well. The "ruling body" of the Bus decided to overturn its own convention bylaws and ultimately did not take on-board a couple of the "top 10" issues as determined by the attendees. Rule by executive fiat is definitely the worst system of all...so excuse me if I'm biased against the "colored dot" solution!

Anne: What places do you have in mind that have tried IRV and then abandoned it? Australia has been using it for 75 years.

Everyone else: Please send your comments to your state legislator and your newspaper editor. The number of people reading this thread is dwindling as it gets longer.

Oh, beaut, let's look at Australia's success. Yeah, 75 years later and they're still spreading vegemite on their crumpets and playing cricket and eating kangaroo meat. Crikey, if Australia is what we would look like if we have IRV, they can keep it, mate.