Washington kind of has this for federal offices (other than president) and state offices. The top two winners of the primary make it to the final election, regardless of party affiliation. We had one state race between a Democrat and a Libertarian, and my congressional district had a choice between two Democrats (though one was 'normal' and the other was a self-proclaimed socialist).

For basically all offices but POTUS that's what California effectively has. Except it is implemented badly as a "top-two primary" system. This virtually guarantees that most partisan elections will exclude anyone outside the two biggest parties. It is not very healthy.

Or outside the one biggest party. Jerry Brown recently vetoed an STV elections bill, calling it "too confusing for voters". Having met some actual voters, I'm growing more inclined to agree with him, but STV has worked very well in my university's elections for more than a decade.

I don't much like STV either; not because it isn't fair, but because it's almost impossible to do without using computers.

We need to keep computers out of elections.

STV has been around since the 19th century, and in use in Australia and Ireland since 1918 and 1921 respectively. At no point have computers been involved.

Although it is simple to perform manually, the iterative nature of counting the votes, eliminating the least popular candidate and distributing their preferences can add extra time to arrive at a result in a tight race. In general, though, the margin between candidates and preference flows are predictable enough that it's rarely significant.

Doesn't STV have the problem of nonmonotonicity, where gaining votes can make a candidate lose?

The algorithm to run an STV election is not that hard to implement. If the government is willing to release the ballot files, then there's no reason that you couldn't conduct your own count. But let's not pretend that we could do massive amounts of tallying without any kind of computer.

If anyone wants to try their hand at implementing the count for a multi-member electorate by PR-STV, the 'formal preferences' files for the Australian 2016 Federal Senate Elections are available here: http://results.aec.gov.au/20499/Website/SenateDownloadsMenu-... (scroll down to "Formal Preferences").

Single-member STV elections are considerably simpler to count.

There's no need to use computers to conduct an STV count. The single-winner variant (variously called "Instant Runoff Voting" "Majority-Preferential Voting", etc.) is especially easy to count by hand. Counting votes in a proportional election under STV is slightly complicated by the transferring of surplus votes, but it can still be done by hand.

Two-round system would be an improvement as well to the FPTP system. It would also fix the spoiler effect and the lack of majority support for winner problems. It's also simpler than RCV.

However, I don't think it reduces negative campaigns like RCV does (though it might improve on the "blue or red" campaigns that exist with FPTP). Another objective may be that the election "costs too much" (basically double or more), although I don't think this is a real objective. Democracy costs money. Deal with it. But I could see the objection gaining traction with some politicians, for the same reason the removal of voting polls gained traction in states with budget deficits.

I'd be content with either RCV or two-round system.

California actually now has a two round system for many legislative (state and federal) and other elections, with the "primary" being the open-field first round and the "general election" being the final vote between the top two.

I've seen lots of complaints that it "eliminates choice", because you have elections where the "general election" options are both Democrats (most notably, this year's U.S. Senate race.)

RCV is better than the Two-Round system, which is an expensive and time-consuming way of doing Supplementary Vote. Supplementary Vote is a system we use in the UK to elect mayors and Police & Crime Commissioners, and is a form of RCV in which voters are limited to only two preferences.

Not certain. http://scorevoting.net/Honest Runoff.html

The address should be http://scorevoting.net/HonestRunoff.html (without the space)

Perhaps not certain, and I do agree that there's far more to be gained by replacing plurality voting by either, but I think that page has some mistakes.

> IRV leads to stifling 2-party domination

I suspect that's because of poor voter information. You can't address every political problem effectively with a choice of voting system.

> with IRV, that one round is more complicated and it cannot be done on ordinary "dumb totalizing" voting machines ... > IRV is more complicated for both voters and talliers.

I find IRV simpler as a voter because of the reduced need for tactical voting in the first round.

IRV isn't complicated to count by hand. (In my country, vested interests said we couldn't afford the necessary voting machines, but we already conduct hand counts suitable for IRV.) You simply separate the ballot papers by candidate and count each candidate's vote, as we do under plurality voting. If no candidate has a majority, eliminate the last candidate, redistribute that candidate's votes to the highest ranked candidate (as marked on each ballot paper) still in the race, count only the redistributed votes and add them to the total from the previous round. Keep going until a candidate has a majority.

> we certainly cannot argue that one system is better than the other under all circumstances.

You have to choose an electoral system before you can know the exact circumstances of an election, but the he main thing is to avoid plurality voting.

A big part of the problem is that the people who are going to approve changing the election system are also often the people who are elected due to the flaws in this system, and would like to hang onto the power these positions give them.

It's as if if the power to accept patches on an OS belonged to people who were profiting from a bunch of zero-day exploits in it, and kept on DMCAing attempts to set up a new repo free of their influence.

But there also hasn't been (as far as I can tell) a significant popular outcry against plurality voting. It seems unfair to blame politicians for resisting a movement that doesn't exist.

There's a chicken-and-egg problem there -- most people don't know there could be any other way to vote.

Two-round systems are actually quite gameable in their naive form.

One version I rather like is a Condorcet election with an optional runoff inside the Smith set, if nontrivial

I agree about the current system, but not about the Two-Round system. It would be more efficient for voters to go to the polls once and choose a first and second preference. Voters could be allowed as many preferences as candidates, which is what RCV does.

As long as the same ballot will be used not just for the runoff but also for the initial election, people have an incentive not to write down their true preferences (this is implied by the https://en.wikipedia.org/wiki/Gibbard%E2%80%93Satterthwaite_... (approximately) which says that as long as there are >2 candidates, tactical voting is incentivized).

The advantage of actually going to the polls twice is that you eliminate tactical voting in the final runoff; people no longer have any incentive to lie about their preferences.

You get to honestly veto the candidate you dislike the most (or back the candidate you want if s/he actually makes the final ballot) in the runoff.

But if anything, the first vote becomes even more tactical potentially in bizarre ways such as aiming to ensure that a candidate sure to make the final ballot runs against the most odious fringe candidate on the final ballot.

Maybe, but I get the impression that overall the addition of a true runoff tends to make things better. For example, this guy compared a number of voting methods using his preferred metric, Bayesian regret; note that when a method and a variant with a runoff added to it are compared, the runoff tends to do slightly better: http://rangevoting.org/StratHonMix.html

In addition, I have no data for this but I expect that a runoff would increase the perception of legitimacy of an electoral result, because while the initial voting procedure that narrows the field to only 2 candidates will occasionally be impacted by counterintuitive results such as spoilers, at least the runoff can always be interpreted in a straightforward manner as "we were choosing between A and B, and most voters preferred A over B".

> Maybe, but I get the impression that overall the addition of a true runoff tends to make things better. For example, this guy compared a number of voting methods using his preferred metric, Bayesian regret; note that when a method and a variant with a runoff added to it are compared, the runoff tends to do slightly better

I suspect that result is caused by voters being misinformed about how IRV works. It would be interesting to see IRV compared to Exhaustive Ballot, which is like the Two-Round system, but with up to n-1 rounds for n candidates. The last-placed candidate is eliminated each round until there are a majority of votes for one candidate, just like IRV.

In practice it rather works the other way: the fact that your ballot will also be used for the final runoff reduces the incentive to tactically vote, because there's a greater chance your vote will backfire.