Non-majority Democracy

[A18]

How would you end the problem of 'vote splitting,' or more specifically, things like:

22% Guy A
15% Guy B
13% Guy C
11% Guy D
09% Guy E
08% Guy F
07% Guy G
06% Guy H
05% Guy I
04% Guy J

Guy A wins with the support of only 22% of the electorate.

[Gabu]

I personally think the best option is the instant runoff voting, where you can list candidates in order of preference.  Once all votes are in, the candidates with the least amount of votes get disqualified and their votes get redistributed to the voters' second choices, then third choices, etc. until someone has more than 50% of the vote.  It's a lot more complicated that the "one vote" method, but I think the extra time would be worth it to make sure that as many people are adequately represented by those in power as possible.

[A18]

Condorcet is the best method if you believe strongly in majority rule. If anyone's interested: http://www.electionmethods.org/Condorcet.htm

I'm split between approval voting and Condorcet. I think I would go with 'approval Condorcet.'

You rank the names of the candidates in order that you prefer them. You can put two candidates in the same rank if you want.

Then you mark each candidate you approve of.

If there's a Condorcet winner, he's the next senator/representative/delegate/whatever. If not, whoever has the highest approval rating wins.

[cwelsch]

Condorcet is incredibly complex for most people.  Arun-off would be the best way to handle a situation like the one above, either IRV or using a two- or three-step process.

[A18]

True, but you don't have to know Condorcet in order to vote using it. You just rank the candidates and check the ones you approve.

[Platypus]

In Australia, you need the support of 50%+1 of the total electorate to win. Not of the total voters, but the electorate. basically, you need to represent the majority to represent anyone.

Compulsory Preferential voting is the only system cutrrently in use anywhere in the world where you need to be supported by the majority to be elected. Bush only won a quarter of the possible vote, for example.

[muon2]

I'm not a fan of either Condorcet or IRV. One main reason is that the current system favors direct election of the government. Let me explain.

In any ranking system, or in a parlimentary system with more than two parties capable of winning seats, the voter tends to vote on ideological grounds for the candidate that best matches their beliefs. With more than two parties electing candidates, some sort of governing coalition must be formed. The formation of the coalition has only minimal input from the voter, usually by parties preannouncing likely alliances.

In a plurality system, a voter can better affect the formation of the government. Essentially the two major parties have already assembled their coalitions, and the voter is cast a vote for their prefered coalition. The parties do realign periodically, and that represents new coalitions forming to govern. Parlimentary procedure assumes that there are exactly two factions representing proponents and opponents, it does not very well handle meetings with more than two strong factions, but forces factions to group into two sides.

I do understand the inherent danger of a candidate elected with a substantial minority of the vote. If I had to modify the system, I would prefer a second runoff election if no candidate reaches a sufficient vote in the first round. This can also be done with open primaries, with a general election only occuring if no candidate gets 50%+1 in the primary. Some of our local non-partisan races use that method.

[Aexoden]

For an office like President, where there's only a single winner, there's no worry about needing to form a coalition. In this kind of situation, any kind of Condorcet system is a good way to find the best candidate.  If there's no Condorcet winner (without access to large numbers of ranked ballots to simulate, it's difficult to say how often this occurs, but I think it's less common than you'd expect), you're faced with the problem of choosing a winner. Approval, as Phillip says, is one way to solve it. Another is using the Ranked Pairs or Beatpath variation on Condorcet, which are generally seen as the best alternatives. In smaller elections especially, it's still possible to run into a problem where there's a tie, but the odds of this would be incredibly small, and it's no worse a problem than a tie during a plurality election.

For legislatures, it's a different issue altogether. If you're for proportional representation, you may have an open list or closed list system. In this case, the parties may have to form coalitions if they want to get anything done (whether or not there should really be that much to get done is another question). My own take would be that people would stick to their principles and propose legislation and vote on legislation according to those principles. Alternatively, a proportional system like STV or CPO-STV (which is Condorcet and STV put together, essentially), would not require the existence of parties, but as parties are valuable for name recognition and financial purposes, I doubt they'd ever really cease to exist. But at least this way, people are choosing their preferred candidates instead of a blanket selection for a party. With the number of "wings" in each party, this is certainly a good thing.

[A18]

Well, the nice thing about American politics is that ALL of our elections are single winner. For example, the House has 435 representatives, but your district only votes in one race for one seat.

My ideal voting system would be computerized with a verifiable paper receipt. Let's say there are five candidates, and you rank them in this order:

--Level 1--
The Republican

--Level 2--
The Libertarian
The Constitutionalist

--Level 3--
The Democrat

--Level 4--
The Green

You hit OK and then you pick your 'approval break' - 1|2   2|3   3|4

The number before the pipestream are the candidates you approve of, and the ones below it are the ones you disapprove of.

As long as there's a Condorcet winner, he wins. Otherwise, add up the approval ratings and see who comes out on top.

I don't think any voting system is perfect, but neither is any other part of government.

[Platypus]

Australia does have a coalition government at the moment, but the two parties have been in a stable coalition for decades. Bascially, the National party only runs in the country, and only in QLD, NSW and VIC, and the Liberals run in the cities and in rural seats in the other states.

Basically, we have a two party system where about 20% of the primary vote goes to mior parties. Of our 150 federal lower house MPs, only 4 aren't in the coalition or opposition. Three are independents, 1 is a green. It's a different state of affairs in the senate, because that uses proportianal preferential voting, so it is more likely that minor parties will be elected to the senate.

[Fmr. Gov. NickG]

In general, I support multi-member districts with membership decided by proportional representation or plural voting.  I don't really support the direct election of the president; I would prefer a parliamentary system.

If we must have elections when a single person must be election, I guess I support IRV (preferential voting) over other systems.  My impression of Cordorcet is that is hugely advantages centrist candidates over more ideological candidates, even when the centrists have no independent base of support.  I think it is healthy in a democracy to cyclically give opposing ideologies the opportunity to govern, so that citizens have a chance to judge the actuall effects of proposed policies once enacted.  We should not be forces to always accept the agenda of the median voter without any opportunity to experiment.

I'm wary of approval voting simple because I don't really understand it.  I understand how the system works, but I don't really understand what this means for how I should personally cast my vote in various situations.  For some reason, asking people to quantify their support for various candidates makes me much more uneasy than asking people to rank them.

[A18]

Condorcet is not at all advantagous to centrists at all. If you prefer a more extreme candidate, you rank him first.

Remember, Condorcet essentially matches up each candidate in a series of one-on-one races. If a person doesn't win every race with a majority, there is no Condorcet winner.

Proportional representation does not take into account geographic rights at all. There could not be a more terrible, factionary voting system.

[Starbucks Union Thug HokeyPuck]

I think we should have another wing of the exectutive branch that is like Parliament.  You win 44%, you get 44% of the seats.  You win 1%, you get 1% of the seats.  Third parties are completely shut out in our current system.  Party with the highest percentage gets the Presidency.  What exactly this Parliament would do?   I dunno, so let's stick with what we got, it works. 

[A18]

The question is about ways to improve our voting system, not to restructure our political process

[Fmr. Gov. NickG]

Condorcet is not at all advantagous to centrists at all. If you prefer a more extreme candidate, you rank him first.

Remember, Condorcet essentially matches up each candidate in a series of one-on-one races. If a person doesn't win every race with a majority, there is no Condorcet winner.

Proportional representation does not take into account geographic rights at all. There could not be a more terrible, factionary voting system.

Condorcet voting helps centrist candidates because it is completely determined by pairwise match-ups.  Assuming a one-dimensional ideological scale, the candidate who is most centrist will beat every other candidate head-to-head, and will be the Condorcet winner.   It doesn't even matter if this centrist candidate wins a single first-place vote.  For instance, in 1992, Ross Perot probably would have won under Condorcet voting.   Candidates on the extremes won't beat any other candidate head-to-head unless they are ranked first by an absolute majority.

Instant Run-Off voting helps centrist candidates somewhat, but only if they have enough initial support to make the final-run off; I think this is a good compromise.

[A18]

Why would a centrist beat every other candidate head to head?

If so, it's because everyone likes him more than any of the other candidates. Sounds fair to me.

[Fmr. Gov. NickG]

Why would a centrist beat every other candidate head to head?

If so, it's because everyone likes him more than any of the other candidates. Sounds fair to me.

Exactly, the centrist candidate beats all the right-wing candidates by getting the votes of the left-wing voters and the centrists, and beats all the left-wing candidates by getting the votes of the right-wing voters and the centrists.  Just like how Perot might have beaten both Clinton and Bush head-to-head.

Whether a centrist candidate with a very low base of support (in terms of 1st places votes) deserves to win is an open question.  Condorcet does not require that a candidate have any first-place votes in order to win...so a a candidate who no one really likes can win just by being inoffensive and middle-of-the-road.  Plurality voting take the opposite extreme, allowing a candidate in a crowded field to win with a small but intense base of support, even if he is hated by everyone else.

I think IRV provides a good middle-ground in this respect.  Although Condorcet may provide the best consensus winner in a given election, I think over the course of several elections, it is best that a voting system allow candidates with a wide range of ideologies to have a shot.  Over time, Condorcet would probably end up giving entrenched power to the party that best represents the median voter.

[Aexoden]

If a candidate is able to get a majority of first-place votes, however (which I would see as the minimum support needed to really be worth anything), they will be elected, so it doesn't necessarily hurt extreme candidates with strong support.

However, it will tend to gravitate toward a more central candidate, though the exact effects or results are hard to gauge. The general concept is that with an extreme candidate, half the people would be very happy, and half would be not at all happy, while with a centrist candidate, everyone would be somewhat happy. This seems the most desirable result to me, as opposed to one extreme majority completely neglecting the minority.

Just what the results would be in a large-scale election, especially since everyone doesn't know enough to rank all the candidates, we may not know until we have such data.

[A18]

Even the right-wingers like him more than the right-wing candidates, then. Same with the left.

So who cares? Everyone's happy.

[willhsmit]

>Even the right-wingers like him more than the right-wing candidates, then. Same with the left.

How do you get that assertion? From what I read, it requires that the overall electorate prefer the centrist to the right-winger, and the overall electorate prefer the centrist to the left-winger. Where does a victory imply that the right-wing prefer the centrist?

Let's say we have 1000 electors...
499 vote:
1 Bush
2 Perot
3 Clinton

499 vote:
1 Clinton
2 Perot
3 Clinton

2 vote:
1 Perot
2 Bush
3 Clinton (or vice versa)

Perot is the Condorcet winner here. Where do you get the idea that this proves the right-wingers prefer him to Bush?

[A18]

Because they just voted for this centrist over a right-winger.

Yes, majority rule. Any majority can decide it, no matter who the winner is.

[willhsmit]

You made the statement, "the right wingers prefer him to the right wing candidate". No, the right-wingers don't. The left-wingers do, and that decides the election.

I'm not really arguing over which is the best way to run an election, I'm just arguing over the obviously nonsensical statement, the right-wingers prefer the centrist candidate to the right-wing one.

[A18]

The right-wingers either do, or he has a BS argument because centrists decide every election.