Statements like the above are one of the reason we don’t see voting system reform in the US. Every time someone discusses reform, people make over-the-top claims like the above (“almost as bad a plurality voting.”)

To the reader: approval / score voting has its own problems, such as not satisfying the majority criterion (a bigger deal than monotonicity) and encouraging strategic voting, among other problems.

I think the more meta problem is that the people who agree that plurality is bad can't agree on what a better system should be :)

IRV actually encourages strategic voting, see e.g. http://mattbruenig.com/2014/11/15/instant-runoff-voting-and-.... For a concrete example, if in the USA we had an IRV election between Trump, Clinton, and Sanders in 2016, there are some (not unrealistic) scenarios where it's possible that your (honest) choice to vote 1. Sanders 2. Clinton 3. Trump would cause a Trump victory, whereas a strategic vote of 1. Clinton 2. Sanders 3. Trump would cause a Clinton victory.

You can argue that approval voting also can lead to strategic voting, but at least with approval voting you won't accidentally cause your least favorite candidate to win by voting honestly.

You should look at Condorcet methods, which as a group are conceptually similar to IRV but are more effective at capturing the intent of the electorate. For example, the above scenario can't happen with Condorcet methods. (Again the problem there is that there are actually many of them, and you have to agree on which specific one you want. But all are better than IRV).

IRV isn't the way forward, and it's unclear to me why organizations like FairVote don't endorse another method. (They have an article about it here but the reasoning doesn't really make sense to me. http://www.fairvote.org/why-the-condorcet-criterion-is-less-...)

> Statements like the above are one of the reason we don’t see voting system reform in the US. Every time someone discusses reform, people make over-the-top claims like the above (“almost as bad a plurality voting.”)

Bucklin Voting's issue is that it can elect the candidate that is everyone's last choice.

Possible does not mean probable. Bayesian Regret figures show Bucklin performing quite well.

http://ScoreVoting.net/StratHonMix.html

You are interpreting that in the wrong way. Just quoting a webpage doesn't lend credence. The methodology is what should be investigated. It's easy to visualize a probable scenario that breaks Bucklin Voting in this political climate.

There are ten candidates. Candidate J is least liked by everyone. Nobody else has a majority until the final preference votes are counted. At that time, the entries are tremendously skewed towards Candidate J, and he wins with the clear majority.

It is a mistake to focus on individual properties.

http://scorevoting.net/PropDiatribe.html

> It is a mistake to focus on individual properties.

That doesn't apply. It is important to filter out voting systems that degenerate into the opposite of their goals. I don't understand why you say that is a mistake. This has nothing to do with a comparison but removing a poor voting system that does not meet the criteria.

No, the reason we don't see voting reform is because there's widespread ignorance of the significance of voting methods. I've spoken to Scott Wiener, Cory Booker, and all manner of activist and politician about this issue, and people generally have a very hard time imagining how this could make a big difference. I point out that we've done exit polls where Approval Voting literally reverses the outcome (i.e. Maine 2014 gubernatorial), and still, they just can't seem to accept it.

At one time, two dozen US cities used STV, and at another time nearly two dozen used Bucklin. And all of them but Cambridge (STV) reverted to Plurality.

So it's just tough, because people are tough.

There's some truth to the statement about IRV. If people are strategic, it essentially degrades to Plurality Voting. http://scorevoting.net/TarrIrv.html

That's not to say that 100% of voters actually are strategic with IRV, but strategic voting appears to have caused IRV to maintain duopoly in countries like Australia—which is pretty undesirable IMO. http://scorevoting.net/AustralianPol.html

Interestingly, polls show that a majority of Australians want to replace STV/IRV with Plurality. http://scorevoting.net/WhatVotersWant

As for these specific criteria, let's be clear from the start that criteria are a very dumb way to assess voting methods, kind of like comparing race cars by comparing horsepower and drag instead of just putting them through timed trials. http://scorevoting.net/PropDiatribe.html

The answer there is Bayesian Regret, which shows Approval Voting performing exceptionally well, and IRV doing worst of the five commonly discussed alternatives. http://scorevoting.net/BayRegsFig.html

Non-monotonicity is an actual problem—it results in a mathematical proof that the voting system has elected the wrong candidate in at least one of two "mirror image" scenarios.

Failing the majority criterion is NOT a problem, in a technical mathematical sense. One of the most elementary facts of social choice theory is that it's possible for a group to prefer X over Y even though a majority of its members prefer Y to X. Mathematical proof here:

http://scorevoting.net/XYvote.html http://scorevoting.net/CondorcetCycles

Though for practical purposes this is all irrelevant, as it would be astronomically unlikely for a majority-preferred candidate to lose with Approval Voting. There is even a theorem that Approval Voting will elect Condorcet winners under rational voter strategy.

http://scorevoting.net/AppCW

With a good voting system, we'd expect/hope to see a competitive race with many candidates, so even having a majority favorite would be the real problem. Without that, the concept of a "majority winner" is nonsensical, yet you still see IRV proponents touting their system's ability to "guarantee a majority winner". I profile the absurdity of this claim here.

https://sites.google.com/a/electology.org/www/utilitarian-ma...