Range Voting

[kevin_standlee]

The folks advocating Range Voting contacted WSFS (actually, the WSFS webmaster, sfrose) lobbying WSFS to change its voting system from the Instant Runoff Voting system we currently use for site selection and the Hugo Awards. Sharon told them how our rules work and suggested that if they want to change them, they come to WSFS business meetings and propose and debate the changes there, like all other rule changes. The advocate's response, in my opinion, amounted to, "Our proposal is so obviously Right that we shouldn't have to do all that hard, expensive work. You should change your rules because we tell you to do so."

I often tell people who come to me with rules-change proposals, "If you think it's worthwhile, come and submit it yourself. I'll help you with all of the technicalities to the best of my ability, but you have to make your own case, lobby people yourself, and get the votes by convincing people." Most of the time, this discourages them -- democracy is hard work! But sometimes we get people who are willing to work and debate, and sometimes we even get workable changes and improvements.

WSFS rules are intentionally designed to be resistant to change; however, they can be changed if people work hard enough at it. But it's not enough to just lobby a Board of Directors or subvert the Chairman; you have to convince the members.

Comments

[thebrokenladder.livejournal.com]

I have a math degree (in fact, we're both in the same general field, aren't we?), and I'm mopping my brains off the floor too. ;)

In what respect? You mean to say that the material is too complicated? Too boring? What exactly is your complaint?

Not to mention, this "social utility efficiency" catchphrase they keep bandying about in their comments here strikes me as so much Boardroom Bingo.

No, it's a measure of your expected value, in the "currency" of satisfaction, with each voting method. A higher social utility efficiency means YOU, Joe Voter, will tend to be happier with the results of elections using Range Voting. Compared to IRV, the effect is enormous. Going from IRV to Range Voting gives you almost as big an increase in your expected satisfaction, as going from random selection to IRV in the first place.

I even took a look at their example, and all I see is that a hypothetical candidate who was left off of three ballots and named last on two others ends up winning. I hardly think a system where a candidate wins whom more than half of the voters absolutely don't want is an effective system.

Leaving someone off the ballot means that you aren't affecting his average, and is generally only done by voters when they don't know enough about that candidate to make an informed decision. In this example, Amy and Bob were extremely disliked by 3 people, whereas Cal was only strongly disliked by two people, and got an almost perfect score from three others, and a very high score from the other. So the most people are the most happy with Cal. The X voters chose to trust the opinions of more informed voters about a candidate they knew little about; but that's their choice, they could have strategically chosen to give Cal a 0 if they wanted.

Contrary to your intuition, this actually is the most effective system, because the paradoxes that arise using other methods tend to be MUCH worse. Look at this IRV election for example (IRV is the method you currently use), where the four voting blocs sit on a left-to-right axis as follows:

Leftist Centrist Rightist
Dean Gore Bush
1
2
3
4

21 Bush > Gore > Dean
10 Gore > Bush > Dean
10 Gore > Dean > Bush
20 Dean > Gore > Bush

With IRV, Gore is eliminated, and Bush beats Dean 31-30. But wait! Gore is preferred to Bush 40-21 - a bigger percentage than any landslide election in our history. IRV picks the wrong winner in this scenario. As a result, the utility efficiency of IRV is significantly lower than that produced by Range Voting.

You can do simulations for yourself if you like, and see which methods tend to leave voters most satisfied. What you'll find is that Range Voting is a very large improvement over IRV, even if we can find special scenarios where Range Voting doesn't seem to pick the winner that you intuitively feel should win the election. On the whole Range Voting picks better winners. As a result, YOU will be happier with the results of elections if you use Range Voting. Choosing NOT to use it is just shooting yourself in the foot. It's like choosing to pick the winner at random instead of holding elections.

Look at this: http://spreadsheets.google.com/pub?key=pXPf6D8HwIWncwYJKKb4CcQ

You can easily set up a spreadsheet like that, and plug in lots of random utility values, and calculate the utility efficiency of plurality, strategic plurality, honest/scaled Range Voting, IRV, strategic IRV (same general strategy as with plurality), etc. etc. You can see the results for yourself. There will be cases when Range Voting picks a worse winner than plurality, but more often than not, it will be vice versa. Range Voting will simply decimate the other methods. Again, I encourage you to test this for yourself. Don't take my word for it. Practice the scientific method.

Clay

[thebrokenladder.livejournal.com]

Ugh...now I see why I have to use the pre element:

Leftist               Centrist                 Rightist
Dean                  Gore                                  Bush
                                                 1
                                         2
                        3
    4

[boywhocantsayno.livejournal.com]

Even with the corrected HTML, I'm not sure I understand this example; that 2nd place vote doesn't seem to be for anything, even "none of the above."

21 Bush > Gore > Dean
10 Gore > Bush > Dean
10 Gore > Dean > Bush
20 Dean > Gore > Bush

With IRV, Gore is eliminated, and Bush beats Dean 31-30. But wait! Gore is preferred to Bush 40-21 - a bigger percentage than any landslide election in our history. IRV picks the wrong winner in this scenario. As a result, the utility efficiency of IRV is significantly lower than that produced by Range Voting.

Actually, under our Constitution, if those were the results than both Gore and Dean would be eliminated after the first ballot, and Bush would be declared the winner 21-20-20:

"Votes shall first be tallied by the voter's first choices. If no majority is then obtained, the candidate who places last in the initial tallying shall be eliminated and the ballots listing it as first choice shall be redistributed on the basis of those ballots' second choices. This process shall be repeated until a majority-vote winner is obtained. If two or more candidates are tied for elimination during this process, the candidate that received fewer first-place votes shall be eliminated. If they are still tied, all the tied candidates shall be eliminated together."

Can I have another example?

[thebrokenladder.livejournal.com]

Yes, I stupidly had Gore and Dean tie here. Change it to percentages as

34
17
15
34

And then you should get the point.

[boywhocantsayno.livejournal.com]

In what respect? You mean to say that the material is too complicated? Too boring? What exactly is your complaint?

Poorly presented, perhaps. Upon a second reading, it looks like a fairly simple system, but you're making it sound much more complex than it needs to be.

I still have a major problem with it: the points you award to a given candidate can be all over the map. A voter who votes "candidate A=9, B=5, C=1" has more weight given to his vote than a voter who opts for "A=3, B=2, C=1" when they've ranked the candidates in the exact same order. This may be a good way of rating a book or a movie for the purpose of telling someone how much you enjoyed it, but it's not good for choosing a convention site or giving out awards when you're comparing two or more candidates directly against each other.

This isn't Olympic figure skating, where every candidate starts out with a score of 6.0 and points are deducted for turning a triple Axel into a double.

Oh, and I also take issue with your example of bees. Simply screaming the loudest is no way to make a democratic decision.

[thebrokenladder.livejournal.com]

Poorly presented, perhaps. Upon a second reading, it looks like a fairly simple system, but you're making it sound much more complex than it needs to be.

I presented it as "Score all the candidates and elect the one who has the highest average." How does that make it sound complicated?

It is important to differentiate between the complexity of the actual voting method, and the complexity of understanding why the voting method is good. In other words, explaining to people what monotonicity and independence of irrelevant alternatives are. Explaining social utility efficiency...that can be frustrating and difficult. But that has nothing to do with how complicated the voting method itself is. When people say that Range Voting would fail under strategic voting, I have to pull out social utility efficiency calculations to prove to them, "No, you're wrong, it wouldn't" and "stop expecting your intuition to have any place in such a deeply mathematical discussion".

A voter who votes "candidate A=9, B=5, C=1" has more weight given to his vote than a voter who opts for "A=3, B=2, C=1" when they've ranked the candidates in the exact same order.

This is a benefit of Range Voting. Say a third voter came in and scored A=0, B=6. B would be the logical winner, even though A is the first choice of two voters. That is, a switch from B to A would only affect voter 2 a tiny amount, and take voter 1 from being moderately happy to being very happy. But voter 3 would go from relative happiness to misery. I don't think that's a good situation. But if voters don't agree, they are free to maximize their scores. In fact, I suggest they do - because what meaning do any scores have if you do not first give your least and most favorite a minimum and maximum score respectively? Without a frame of reference, what do the numbers even mean?

So, in short, don't blame the voting system when voters do silly things. This reminds me of the 97,488 Floridians who voted for Nader in 2000, only to help secure the victory for Bush. Should we have taken Nader off the ballot, to protect them from their own silliness?

This may be a good way of rating a book or a movie for the purpose of telling someone how much you enjoyed it, but it's not good for choosing a convention site or giving out awards when you're comparing two or more candidates directly against each other.

On the contrary, it's the best of the common single-winner voting methods, based on extensive calculations of social utility efficiency (http://RangeVoting.org/vsr.html). That is, by using Range Voting, you will be greatly more satsified with election results than you would be with IRV. The effect is almost as large as the one caused by going from random selection, to IRV elections like you have now.

This isn't Olympic figure skating, where every candidate starts out with a score of 6.0 and points are deducted for turning a triple Axel into a double.

I don't recall ever saying that it was. It has been compared more to gymnastics scoring, but even then that's just to give people the concept in a simple recognizable scenario. Maybe "like rating books on Amazon" would be better for certain audiences.

[thebrokenladder.livejournal.com]

34% McCain > Gore > Edwards
17% Gore > McCain > Edwards
15% Gore > Edwards > McCain
34% Edwards > Gore > McCain

McCain wins with IRV, even though 66% of voters prefer Gore to both McCain and Edwards. IRV picks the wrong winner. Any election method can exhibit this kind of phenomenon, but with IRV, it's negative impact on social utility is significantly worse.