Slashdot Mirror

http://rangevoting.org/VenHist.html describes how the republic of Venice elected its Doge, the highest ranking official. In some of the phases of the process, it is similar to the random assignment of music to one of Salganik's eight artificial "worlds", which had to work with a much larger population. (Note that Venice had a population of 150,000 in 1630, roughly one two-thousandth the current population of the US.)

Nominees were often chosen by committees, who in turn were selected by a hopefully-incorruptible random process (involving selecting balls from urns) then the election for that position was among those who had been nominated. By having multiple stages of both random and election processes the Venetians tried to make the system incorruptible (thanks to the randomness) but also striving for maximum quality (due to the democratic electing-the-best processes).

Thus the process for electing the Doge, as of 1268 (when it was employed for the election of Lorenzo Tiepolo), had reached this amazing almost-final form [Lane p.111; also described by Lines p.156]:

      1. Choose 30 of the Great Council members (of whom there were 1000-to-1500, typically; all male) by a random process;
      2. Reduce them to 9 by random processes;
      3. The 9 name 40 nominees;
      4. The 40 are reduced to 12 by a random process;
      5. the 12 name 25 nominees;
      6. Reduce them to 9 by random processes;
      7. The 9 name 45 nominees;
      8. Reduce them to 11 by random processes;
      9. The 11 named 41 (all of whom had to be ageâ¥40 years);
    10. The 41 elected the Doge (from among nominees they chose; any of the 41 could write a name on a slip of paper, and from then onward, that name was a candidate) by range3 voting!
    11. This choice theoretically was subject to approval or veto by the mass of the people (assembly) but I am unaware of any instance in which that veto was exercised. This perhaps meant this step was a mere formality with the People not really having any power. But another interpretation is that the threat of a veto kept the Grand Council honest in its choice they refused to risk the embarrassment of a veto.

In this process, only the penultimate step â" the election â" "really mattered" â" the rest was mainly intended to make the identity of the 41 unpredictable hence making the process (hopefully) uncorruptible. The 41, during their deliberations, were sequestered rather like the juries in modern-day big-time criminal cases. This again was presumably intended to insulate them from corruption.

Perhaps it's time to start a petition for the Range Voting method to become the standard for all US voting that will put a person into a political office? Perhaps then we would see a more honest accounting of the opinions of those in the USA.

If we were to be able to actually vote for our top preferences, instead of just one option, then I'd finally be able to vote my real opinion, and not have to have my top vote be either a red or a blue. Instead of "throwing away" my vote, I could vote my conscience, giving highest marks to my preferred candidate, but still having the option of giving high marks to a good enough alternative. What this would mean is that I can give my preferred candidate a score of 100, a good alternate candidate a score of 85, score my two next favorites that I equally prefer over all others as scores of 50 for both, and score those I vehemently disagree with a score of zero. This system truly gives me my real vote (no wasted votes either), and gives me a better chance at being truly represented. The current system forces me to choose the lesser of two evils and only truly gives me 2 real choices.

<soapbox>
I want my true representation! How long will we continue down this road of taxation without representation. 35% of my income goes to the government, and the system does not truly give me a way to be represented (in fact, it most often only gives 35-45% of voters representation by their chosen candidate. The rest of us are not represented).
</soapbox>

Sorry for the rant, but I feel like it's time for the "tree of liberty" to be refreshed, so to speak, "by the blood of patriots and tyrants." Thomas Jefferson had better insight that he could ever have imagined.

We've gone stale as a nation, and we are now like a rotting corpse.

It sounds good to say that, but how do you actually do it?

Condorcet voting systems, or range voting.

There are lots of voting systems that eliminate the "spoiler" effect that you're complaining about; the problem is that too few people seem to know about them, and of those that know, too few care enough to actually get anything changed.

And, of course, the two major parties are quite happy with their little duopoly, so the only way to fix things is by citizen initiatives to change state constitutions.

The pointlessness of a two-party system based on false antagonisms and dichotomies. Sadly, there seems to be no hope in sight.

Either approval voting or range voting (aka score votingwould break the forced two-valued dichotomy of the current system.

(In fact, approval voting is just one version of range voting-- in games theory, they are identical).

Single Transferrable Vote is what you want.

A) STV is only good as a multi-winner method, but is rather terrible as a single-winner method (aka IRV). Score Voting (aka Range Voting) is massively better. => http://rangevoting.org/CFERlet.html B) Reweighted Score Voting and Asset voting are simpler and better than STV. http://rangevoting.org/PropRep.html Although it'll take some explaining to some of the populace. So offer alternatives.

At the voting stations, have two lines, a fast track and a slow track.

In the fast track, you get a ballot paper with the usual STV instructions - place a 1 next to your first choice candidate, a 2 in your second choice candidate, and so on until you have no more preferences between the remaining candidates.

In the slow track, a computer screen. It says "Which of these do you want to win?" and a list of candidates. The voter selects one.

Then it says "If that guy doesn't win, which of these would you like to win?" followed by a list of the remaining candidates and a further option "They're all as bad as each other".

Single Transferrable Vote is what you want.

A) STV is only good as a multi-winner method, but is rather terrible as a single-winner method (aka IRV). Score Voting (aka Range Voting) is massively better. => http://rangevoting.org/CFERlet.html B) Reweighted Score Voting and Asset voting are simpler and better than STV. http://rangevoting.org/PropRep.html Although it'll take some explaining to some of the populace. So offer alternatives.

At the voting stations, have two lines, a fast track and a slow track.

In the fast track, you get a ballot paper with the usual STV instructions - place a 1 next to your first choice candidate, a 2 in your second choice candidate, and so on until you have no more preferences between the remaining candidates.

In the slow track, a computer screen. It says "Which of these do you want to win?" and a list of candidates. The voter selects one.

Then it says "If that guy doesn't win, which of these would you like to win?" followed by a list of the remaining candidates and a further option "They're all as bad as each other".

No it wouldn't. http://rangevoting.org/TarrIrv.html

ALL voting methods can fail to elect the most popular candidate. But extensive Bayesian regret calculations show that Score Voting (aka Range Voting) does the best job of any feasible voting method at electing the most popular candidate possible (where "most popular" means "most liked by the most people", not necessarily "the favorite candidate of the most people"). http://rangevoting.org/UniqBest.html http://rangevoting.org/StratHonMix.html

ALL voting methods can fail to elect the most popular candidate. But extensive Bayesian regret calculations show that Score Voting (aka Range Voting) does the best job of any feasible voting method at electing the most popular candidate possible (where "most popular" means "most liked by the most people", not necessarily "the favorite candidate of the most people"). http://rangevoting.org/UniqBest.html http://rangevoting.org/StratHonMix.html

Why did you ignore what I said about the Balkans? Clinton reversed himself in Somalia in the way that many feel Bush should have in Iraq, but Clinton was a major actor in the UN involvement in the Balkans which persists to this day. (Albeit under the purview of the EU.) If Iraq is a quagmire, then Bosnia is a quagmire

I think where we differ is that I see Bosnia as a competently executed action with a successful outcome, not as a botched ongoing quagmire. The Bosnian intervention was well handled, had full support of NATO, and the reasons put forth for doing it were factual and valid. Just as importantly, the costs and the casualties were light, and were spread across multiple countries. According to figures I have in front of me, the only NATO casualties during the war were two Americans killed when their helicopter crashed. Estimates of civilian casualties inflicted by NATO range from 1200 to 5700. The political situation in Bosnia stabilized within months, not years.

Compare that to what happened (and continues to happen!) in Iraq. Compare the casualties and costs there. Examine who did (and didn't) bear the vast majority of the costs in Iraq. And finally, examine the validity of the reasons put forth for invading Iraq, and therefore the justness and morality of the invasion. Iraq comes up short on all counts, IMHO.

But Hillary Clinton and John Kerry authorized Iraq. None of this seems to add up to there being a clear difference between the parties.

True, which is one of the main reasons why I didn't support Hillary Clinton in the 2008 primaries. Obama, on the other hand, was against Iraq from the start, and articulated very clear, common-sense reasons as to why Iraq was a bad idea... which is why I supported Obama (and still support him, despite his bad vote on FISA). Again, the parties are made of individuals, not of a monolithic borg collective, and you can only generalize so far before the generalization becomes more misleading than helpful.

Neither party is running an anti-FISA presidential candidate. Done deal. Both parties have anti-FISA members.

Perhaps, but FISA is not the only important issue to me. The Iraq war, the environment, the economy, women's reproductive rights, the role of corporations in our government and society... those are all issues that I care about, and on all of those issues I much prefer the Democratic policies to the Republican ones. So despite my dissappointment on the whole warrantless-wiretapping issue, I still think that our country will be significantly better off under the Democratic Party than with another four years of Republican rule.

To your larger point, the electorate largely gets the candidates we deserve. I think, however, that your attitude that only Democrats and Republicans are worthy of consideration is part of the problem. The only way you can waste your vote is by voting for a candidate you don't want.

I think your perception of my attitude is a little bit off -- it's not that I don't think third-party candidates are worthy of consideration (hell, I was a registered member of the Green Party for a number of years)... it's just that I can see that in a plurality election system, what usually happens is that the perfect becomes the enemy of the good. In an ideal world, we'd use an adequate voting system (e.g. Range Voting or Condorcet Voting) that would properly handle elections with more than two viable candidates, and everybody could always vote for the candidate they liked best and that would lead to good results. However, I have to be realistic and acknowledge that we don't live in that world: we live in a country that uses Plurality Voting, which is indisputably the worst electoral system when it comes to generating the desired result in races with three or more competitive candidates. In a Plurality Voting s

But until the electoral college is abolished, we still have no real vote on certain candidates

Total nonsense. The problem isn't the electoral college, it's that we haven't got range voting -- they're not incompatible. As a matter of fact, given the electoral college, you can lobby your state government and have your state implement range voting as applied to your state's electoral college votes. Getting one state to do something is generally far easier than getting the federal government to do it, and if you manage it then it will be in the news -- in the next election the media will be explaining how XYZ state now has a different voting system and now everybody watching at home can say, why don't we have that?

But first you have to get off your ass and convince your state to use range voting.

We need instant-runoff voting to avoid the whole viability question: http://en.wikipedia.org/wiki/Instant-runoff_voting

Ew, no. IRV is needlessly complex, too easy to game, and swings around unpredictably between "2 big parties 1 little party" to "1 big party 2 little parties", resulting in no actual IRV-using countries with more than 2 parties/blocs/coalitions. Condorcet is nicer, but suffers from the "Dark Horse" problem.

Approval and Range (the former being a special case of the latter) are even better, and both sidestep Arrow's Impossibility Theorem by not being ranked ballots.

We need instant-runoff voting to avoid the whole viability question: http://en.wikipedia.org/wiki/Instant-runoff_voting

Ew, no. IRV is needlessly complex, too easy to game, and swings around unpredictably between "2 big parties 1 little party" to "1 big party 2 little parties", resulting in no actual IRV-using countries with more than 2 parties/blocs/coalitions. Condorcet is nicer, but suffers from the "Dark Horse" problem.

Approval and Range (the former being a special case of the latter) are even better, and both sidestep Arrow's Impossibility Theorem by not being ranked ballots.

We need instant-runoff voting to avoid the whole viability question: http://en.wikipedia.org/wiki/Instant-runoff_voting

Ew, no. IRV is needlessly complex, too easy to game, and swings around unpredictably between "2 big parties 1 little party" to "1 big party 2 little parties", resulting in no actual IRV-using countries with more than 2 parties/blocs/coalitions. Condorcet is nicer, but suffers from the "Dark Horse" problem.

Approval and Range (the former being a special case of the latter) are even better, and both sidestep Arrow's Impossibility Theorem by not being ranked ballots.

Ask Nader voters how that worked out for them in 2000. In an ideal world, your advice would be good advice. But we in the US are blessed with an incredibly flawed electoral system where strategic concerns routinely trump the voter's sincere preferences in determining the result. Asking the voters to willingly ignore that fact and shoot themselves in the foot is asking too much, and even if they did follow your advice it would only make things worse for all of us (again, see Florida 2000 for an example).

The proper way to resolve this ugly problem is to update our electoral system to something that works (Range Voting or Condorcet are both good options), but that is a difficult task and it's not going to happen in the near term. So we keep pushing for that, but in the meantime we have to make the best of the system we have, and that means taking its flaws into account when we vote.

It's an interesting idea, but I think Approval and Range voting both do it better. (Approval: vote for every candidate who's "good enough", then add up the votes the same way as the current system. Range: vote each candidate on a scale, e.g. 0 to 99, add up the scores, divide by the number of voters.)

I think your idea falls down in the two corner cases that any voting system ought to account for: third party candidates whom the voter doesn't know, and write-in candidates.

Let's consider unknown third parties first. Suppose you look at the ballot, and you've got a good idea of how you're voting for the Democrats, Republicans, Libertarians, Greens, and so on. Then you see a candidate from, say, the Cheese Party. Huh? In your system, you can either vote for him (+1), or vote against him (-1). If you're like most people, you vote against him since you don't know if he'd be any good. You certainly won't vote for him. So even if the candidate was actually pretty sensible, the Cheese Party loses by a landslide, nobody ever looks into the Cheese Party, and the Cheese Party loses by a landslide in the next election, too. (Approval voting suffers from the same problem here.)

Range voting lets voters leave a choice blank. "Blank" means "I don't have enough information to cast a vote, so I'll leave this choice to the more informed voters". (Very useful in states where you vote on judges, but judges' party affiliations are forbidden on the ballot, and there are so many races that it's impossible to research them all.) Now, Range Voting as normally implemented has a quorum requirement, so the Cheese Party still loses since few people had heard of them, but they score really well (among people who did vote for them), they make it into the newspaper, more people hear about the Cheese Party platform and what they stand for, and next time around more people give a score for their candidate. Maybe someday they win.

Now, let's consider write-in candidates. Your system could either (A) forbid write-ins entirely, (B) automatically treat un-written-in ballots as "0", or (C) automatically treat un-written-in ballots as "-1". Assuming write-in candidates are a good thing, A is obviously a bad choice, and C results in the same problem mentioned previously for third parties. B, however, is even more dangerous: in your system, the final tallies for any winning candidate will almost always be small positive integers; even a small write-in campaign can completely topple the preferences of a much larger number of voters. We'll assume that C is the system actually used. (Approval is most similar to C.)

Range voting, OTOH, can handle write-in candidates without blinking, since the system is already set up to gracefully handle "blank" votes on unknown candidates. If the write-in campaign is too small, they make a good showing but are disqualified by the quorum rule. If the write-in campaign is big enough, maybe they win. For a third party that's just starting out, being off the ballot isn't much different from being on the ballot, which means that all the petty politics of ballot qualification become irrelevant.

It's an interesting idea, but I think Approval and Range voting both do it better. (Approval: vote for every candidate who's "good enough", then add up the votes the same way as the current system. Range: vote each candidate on a scale, e.g. 0 to 99, add up the scores, divide by the number of voters.)

I think your idea falls down in the two corner cases that any voting system ought to account for: third party candidates whom the voter doesn't know, and write-in candidates.

Let's consider unknown third parties first. Suppose you look at the ballot, and you've got a good idea of how you're voting for the Democrats, Republicans, Libertarians, Greens, and so on. Then you see a candidate from, say, the Cheese Party. Huh? In your system, you can either vote for him (+1), or vote against him (-1). If you're like most people, you vote against him since you don't know if he'd be any good. You certainly won't vote for him. So even if the candidate was actually pretty sensible, the Cheese Party loses by a landslide, nobody ever looks into the Cheese Party, and the Cheese Party loses by a landslide in the next election, too. (Approval voting suffers from the same problem here.)

Range voting lets voters leave a choice blank. "Blank" means "I don't have enough information to cast a vote, so I'll leave this choice to the more informed voters". (Very useful in states where you vote on judges, but judges' party affiliations are forbidden on the ballot, and there are so many races that it's impossible to research them all.) Now, Range Voting as normally implemented has a quorum requirement, so the Cheese Party still loses since few people had heard of them, but they score really well (among people who did vote for them), they make it into the newspaper, more people hear about the Cheese Party platform and what they stand for, and next time around more people give a score for their candidate. Maybe someday they win.

Now, let's consider write-in candidates. Your system could either (A) forbid write-ins entirely, (B) automatically treat un-written-in ballots as "0", or (C) automatically treat un-written-in ballots as "-1". Assuming write-in candidates are a good thing, A is obviously a bad choice, and C results in the same problem mentioned previously for third parties. B, however, is even more dangerous: in your system, the final tallies for any winning candidate will almost always be small positive integers; even a small write-in campaign can completely topple the preferences of a much larger number of voters. We'll assume that C is the system actually used. (Approval is most similar to C.)

Range voting, OTOH, can handle write-in candidates without blinking, since the system is already set up to gracefully handle "blank" votes on unknown candidates. If the write-in campaign is too small, they make a good showing but are disqualified by the quorum rule. If the write-in campaign is big enough, maybe they win. For a third party that's just starting out, being off the ballot isn't much different from being on the ballot, which means that all the petty politics of ballot qualification become irrelevant.

It's an interesting idea, but I think Approval and Range voting both do it better. (Approval: vote for every candidate who's "good enough", then add up the votes the same way as the current system. Range: vote each candidate on a scale, e.g. 0 to 99, add up the scores, divide by the number of voters.)

I think your idea falls down in the two corner cases that any voting system ought to account for: third party candidates whom the voter doesn't know, and write-in candidates.

Let's consider unknown third parties first. Suppose you look at the ballot, and you've got a good idea of how you're voting for the Democrats, Republicans, Libertarians, Greens, and so on. Then you see a candidate from, say, the Cheese Party. Huh? In your system, you can either vote for him (+1), or vote against him (-1). If you're like most people, you vote against him since you don't know if he'd be any good. You certainly won't vote for him. So even if the candidate was actually pretty sensible, the Cheese Party loses by a landslide, nobody ever looks into the Cheese Party, and the Cheese Party loses by a landslide in the next election, too. (Approval voting suffers from the same problem here.)

Range voting lets voters leave a choice blank. "Blank" means "I don't have enough information to cast a vote, so I'll leave this choice to the more informed voters". (Very useful in states where you vote on judges, but judges' party affiliations are forbidden on the ballot, and there are so many races that it's impossible to research them all.) Now, Range Voting as normally implemented has a quorum requirement, so the Cheese Party still loses since few people had heard of them, but they score really well (among people who did vote for them), they make it into the newspaper, more people hear about the Cheese Party platform and what they stand for, and next time around more people give a score for their candidate. Maybe someday they win.

Now, let's consider write-in candidates. Your system could either (A) forbid write-ins entirely, (B) automatically treat un-written-in ballots as "0", or (C) automatically treat un-written-in ballots as "-1". Assuming write-in candidates are a good thing, A is obviously a bad choice, and C results in the same problem mentioned previously for third parties. B, however, is even more dangerous: in your system, the final tallies for any winning candidate will almost always be small positive integers; even a small write-in campaign can completely topple the preferences of a much larger number of voters. We'll assume that C is the system actually used. (Approval is most similar to C.)

Range voting, OTOH, can handle write-in candidates without blinking, since the system is already set up to gracefully handle "blank" votes on unknown candidates. If the write-in campaign is too small, they make a good showing but are disqualified by the quorum rule. If the write-in campaign is big enough, maybe they win. For a third party that's just starting out, being off the ballot isn't much different from being on the ballot, which means that all the petty politics of ballot qualification become irrelevant.

No, IRV does not eliminate spoilers. The spoiler effect exists in IRV. With IRV there are still wasted votes, except it's much harder to tell which votes will be wasted and what effect your ballot will have; the behaviour of IRV is much more complicated, often pathological, and thus arguably even worse than the current system.

Range or approval voting would be a better option. They truly eliminate the spoiler effect, they are easy to implement using unaltered existing equipment, and have simple, easily understandable behaviour.

I will use my home state of Kansas as a good example of "Gerrymandering." Here is a map of our congressional districts. As you can see, the third district is, by area, the smallest, but it contains an equal amount of people when compared to the other districts, as the Kansas City/Overland Park area is the most populated. It also houses almost all the Kansas Democrats. How would you redistribute the voting districts? If you did it by horizontal or vertical stripes of population distribution, the Democrats would not get a say in many votes, as they would be overpowered by the Republicans populating the rest of the state. In fact, the current solution is the only way Democrats get any say. Should the Democrats get a vote? Kansas has been a red state for a very long time. What I'm saying is there isn't really a good solution to the voting district problem (based on how we currently elect officials, anyway).

I will use my home state of Kansas as a good example of "Gerrymandering." Here is a map of our congressional districts. As you can see, the third district is, by area, the smallest, but it contains an equal amount of people when compared to the other districts, as the Kansas City/Overland Park area is the most populated. It also houses almost all the Kansas Democrats. How would you redistribute the voting districts? If you did it by horizontal or vertical stripes of population distribution, the Democrats would not get a say in many votes, as they would be overpowered by the Republicans populating the rest of the state. In fact, the current solution is the only way Democrats get any say. Should the Democrats get a vote? Kansas has been a red state for a very long time. What I'm saying is there isn't really a good solution to the voting district problem (based on how we currently elect officials, anyway).

Do you seriously not get it, or are you deliberately trying to be obtuse? Of course the people who get elected got more votes, gerrymandering doesn't change the basic electoral system. What it does do is allow politicians to run virtually uncontested in a gerrymandered district. Since you completely ignored the rather obvious example of how this works provided by jsprat (in pictorial form, even!), I'll attempt to explain it again. In the original image, the two parties, green and magenta, are equally distributed in terms of voters (the dots). This would let to some very close elections, ranging from 4 green representatives to 4 magenta and everything in between (we'll say it evens out to 2 and 2). Now look at the gerrymandered example. There are now 3 representatives elected from the magenta party with no contest. See the problem?

If for some reason an elegantly simple example, such as the one in the Wikipedia article, is not sufficient, how about some real world examples? Some of these districts are downright ludicrous. Are you seriously trying to tell me these district lines were drawn in an effort to create fair and unbiased voting districts?

The fact is, lots of people will be honest when using Range Voting, and will not strategically exaggerate into Approval Voting. And that's not a big deal since honest Range Voting is about 91% as effective as strategic Range Voting.

You have to consider that around 1/3 of people weaken their vote power by not registering in a major party in states where you have to if you want to vote in a major party primary. They prefer honesty to power.

You should check out:
http://rangevoting.org/HonStrat.html
http://rangevoting.org/ShExpRes.html
and lots of other stuff on RangeVoting.org
Your arguments are not novel, and they have been rigorously addressed by Range Voting proponents.

The fact is, lots of people will be honest when using Range Voting, and will not strategically exaggerate into Approval Voting. And that's not a big deal since honest Range Voting is about 91% as effective as strategic Range Voting.

You have to consider that around 1/3 of people weaken their vote power by not registering in a major party in states where you have to if you want to vote in a major party primary. They prefer honesty to power.

You should check out:
http://rangevoting.org/HonStrat.html
http://rangevoting.org/ShExpRes.html
and lots of other stuff on RangeVoting.org
Your arguments are not novel, and they have been rigorously addressed by Range Voting proponents.

Ever been to a site that allows people to vote on articles on a scale of 1-10? It rapidly degenerates into everyone either voting 10 or 0 1. Even if everyone does that, then you have "Approval Voting", which is still exceptionally good. 2. We've done a lot of research that indicates A LOT of people will choose to be expressive and use their full range of values instead of min/max, especially since it has about 91% as much effectiveness as perfectly strategizing (which is hard to do since it requires probability analysis). (e.g. You might consider that around 1/3 of voters don't register with a major party, even in areas where that prevents them from voting in the primaries, and thus diminishes their vote power.) http://rangevoting.org/HonStrat.html 3. That extra honesty means much better election outcomes, so Range Voting is much more "effective" than Approval Voting. http://rangevoting.org/ShExpRes.html 4. Range Voting can help with problems like the Burr dilemma, and is experimentally shown to be "easier" in the sense that voters will sometimes have a hard time making a yes/no binary decision for a candidate who is right on the line, but will have an easier time scoring (which is why HotOrNot's creators preferred scores to hot?/not? approval voting -- faster scoring is good for their "business model"). 5. Range Voting has a powerful "nursery effect" that gives fair representation of minor candidates/parties/viewpoints until they become competitive, so a candidate who may get very few approvals with approval voting can maybe get a 35% average, and that can help society gauge itself -- it can work like a windsock to let us get a better sense of the political winds. That's abstract but plausibly very beneficial to society. Approval is definitely simpler, and is a great voting method that we can implement RIGHT NOW with virtually no changes. From there we could upgrade to Range Voting if possible. Hence: http://rangevoting.org/Approval.html Here's a head-to-head comparison/apologetics piece: http://rangevoting.org/AppExec.html

Ever been to a site that allows people to vote on articles on a scale of 1-10? It rapidly degenerates into everyone either voting 10 or 0 1. Even if everyone does that, then you have "Approval Voting", which is still exceptionally good. 2. We've done a lot of research that indicates A LOT of people will choose to be expressive and use their full range of values instead of min/max, especially since it has about 91% as much effectiveness as perfectly strategizing (which is hard to do since it requires probability analysis). (e.g. You might consider that around 1/3 of voters don't register with a major party, even in areas where that prevents them from voting in the primaries, and thus diminishes their vote power.) http://rangevoting.org/HonStrat.html 3. That extra honesty means much better election outcomes, so Range Voting is much more "effective" than Approval Voting. http://rangevoting.org/ShExpRes.html 4. Range Voting can help with problems like the Burr dilemma, and is experimentally shown to be "easier" in the sense that voters will sometimes have a hard time making a yes/no binary decision for a candidate who is right on the line, but will have an easier time scoring (which is why HotOrNot's creators preferred scores to hot?/not? approval voting -- faster scoring is good for their "business model"). 5. Range Voting has a powerful "nursery effect" that gives fair representation of minor candidates/parties/viewpoints until they become competitive, so a candidate who may get very few approvals with approval voting can maybe get a 35% average, and that can help society gauge itself -- it can work like a windsock to let us get a better sense of the political winds. That's abstract but plausibly very beneficial to society. Approval is definitely simpler, and is a great voting method that we can implement RIGHT NOW with virtually no changes. From there we could upgrade to Range Voting if possible. Hence: http://rangevoting.org/Approval.html Here's a head-to-head comparison/apologetics piece: http://rangevoting.org/AppExec.html

Ever been to a site that allows people to vote on articles on a scale of 1-10? It rapidly degenerates into everyone either voting 10 or 0 1. Even if everyone does that, then you have "Approval Voting", which is still exceptionally good. 2. We've done a lot of research that indicates A LOT of people will choose to be expressive and use their full range of values instead of min/max, especially since it has about 91% as much effectiveness as perfectly strategizing (which is hard to do since it requires probability analysis). (e.g. You might consider that around 1/3 of voters don't register with a major party, even in areas where that prevents them from voting in the primaries, and thus diminishes their vote power.) http://rangevoting.org/HonStrat.html 3. That extra honesty means much better election outcomes, so Range Voting is much more "effective" than Approval Voting. http://rangevoting.org/ShExpRes.html 4. Range Voting can help with problems like the Burr dilemma, and is experimentally shown to be "easier" in the sense that voters will sometimes have a hard time making a yes/no binary decision for a candidate who is right on the line, but will have an easier time scoring (which is why HotOrNot's creators preferred scores to hot?/not? approval voting -- faster scoring is good for their "business model"). 5. Range Voting has a powerful "nursery effect" that gives fair representation of minor candidates/parties/viewpoints until they become competitive, so a candidate who may get very few approvals with approval voting can maybe get a 35% average, and that can help society gauge itself -- it can work like a windsock to let us get a better sense of the political winds. That's abstract but plausibly very beneficial to society. Approval is definitely simpler, and is a great voting method that we can implement RIGHT NOW with virtually no changes. From there we could upgrade to Range Voting if possible. Hence: http://rangevoting.org/Approval.html Here's a head-to-head comparison/apologetics piece: http://rangevoting.org/AppExec.html

Ever been to a site that allows people to vote on articles on a scale of 1-10? It rapidly degenerates into everyone either voting 10 or 0 1. Even if everyone does that, then you have "Approval Voting", which is still exceptionally good. 2. We've done a lot of research that indicates A LOT of people will choose to be expressive and use their full range of values instead of min/max, especially since it has about 91% as much effectiveness as perfectly strategizing (which is hard to do since it requires probability analysis). (e.g. You might consider that around 1/3 of voters don't register with a major party, even in areas where that prevents them from voting in the primaries, and thus diminishes their vote power.) http://rangevoting.org/HonStrat.html 3. That extra honesty means much better election outcomes, so Range Voting is much more "effective" than Approval Voting. http://rangevoting.org/ShExpRes.html 4. Range Voting can help with problems like the Burr dilemma, and is experimentally shown to be "easier" in the sense that voters will sometimes have a hard time making a yes/no binary decision for a candidate who is right on the line, but will have an easier time scoring (which is why HotOrNot's creators preferred scores to hot?/not? approval voting -- faster scoring is good for their "business model"). 5. Range Voting has a powerful "nursery effect" that gives fair representation of minor candidates/parties/viewpoints until they become competitive, so a candidate who may get very few approvals with approval voting can maybe get a 35% average, and that can help society gauge itself -- it can work like a windsock to let us get a better sense of the political winds. That's abstract but plausibly very beneficial to society. Approval is definitely simpler, and is a great voting method that we can implement RIGHT NOW with virtually no changes. From there we could upgrade to Range Voting if possible. Hence: http://rangevoting.org/Approval.html Here's a head-to-head comparison/apologetics piece: http://rangevoting.org/AppExec.html

Well, for your Senate you use proportional STV, which is a pretty decent voting method, although rather "archaic" in the sense that simpler methods have come about that are mathematically superior in terms of proportionality and criteria like monotonicity. E.g. Reweighted Range Voting and Asset Voting (the latter was originally invented by Lewis Carroll). But for your House elections you use the single-winner form of STV, called "Instant Runoff Voting", and it is a rather horrible voting method, according to Smith's Bayesian regret studies. Especially when voters are strategic: http://rangevoting.org/StratHonMix.html

Warren D. Smith, the Princeton math Ph.D. who was featured in a recent article on Range Voting, has looked at this issue. http://rangevoting.org/Apportion.html http://rangevoting.org/SplitLR.html

Warren D. Smith, the Princeton math Ph.D. who was featured in a recent article on Range Voting, has looked at this issue. http://rangevoting.org/Apportion.html http://rangevoting.org/SplitLR.html

http://www.rangevoting.org/GerryExamples.html

No affiliation - was just googling up some pictures to support my own (lesser) ideas for simple geometric rules to limit gerrymandering.

Ranked Pairs is far and away the best option. It satisfies the monotonocity criterion. It is very simple to understand, especially from the perspective of a voter. You simply rank who you want to vote for. (Any voter who doesn't like it can simply rank their candidate #1 and leave the rest blank, just as we do now.)

I was a huge fan of Schulze method (another Condorcet variant, and one used in the real world in Debian elections), until RangeVoting.org convinced me today that Favorite Betrayal is an actual problem in any Condorcet method. Not only does Range Voting not have this problem, it also bypasses a lot of the Arrow's Impossibility Theorem baggage (since Arrow only applies to ranked systems, not scored systems) and, even better, is nearly as simple as Plurality or Approval.

(As the RV.org site points out, when you sit down and write code to implement a Condorcet method, any of them, you end up with about twice as much code compared to implementing RV. I've implemented Schulze voting before, so I know that of which they speak. Meanwhile, with RV, you just sum the votes as you would with Plurality or Approval, then divide by the number of voters. The fact that each vote is a ranged score rather than a {0|1} doesn't really change how the code works. Maybe you add a simple quorum check on the final score sums, but that's 2 lines of Perl using sort and grep.)

Plus bee swarms use it to vote on new hive locations, so even though I was skeptical at first about the Bayesian regret metric, RV is a tried and tested voting method in the real world, and bees are considerably less intelligent than even the dumbest voter.

There are two main problems with range voting. One, it makes voting overly complicated (not to the Slashdot crowd, but for the average voter.) Two, it is subject to gaming. Voters will be likely to rate secondary candidates lower than how they really feel in order to increase their top candidate's odds.

Re #1: I'd argue that RV is actually simpler than any ranked ballot. Seven words: Hot or Not, Olympic scoring, IMDb, Netflix. Every Joe-on-the-Street already understands RV. It's one of the three simplest systems possible (after Approval and Plurality). Try out this Range Vote on the US Presidential race — unlike Condorcet, you don't have to stand around hemming and hawing about exactly which order you'd put them in, plus you can actually express "I have no opinion" by leaving a line blank. (And since they threw in all the kitchen sink candidates on that poll, you'll use that option a lot.)

(In Condorcet, your only option for unknown candidates is to lump them all together in one big tie in the middle... but you're still saying "These unknown candidates are definitely worse than any of the ones I like" and "These unknown candidates are definitely better than any of the ones I hate". One of them could be {your undiscovered favorite|the reincarnation of Hitler}, but you've just {hurt|helped} their campaign. RV lets you leave them blank, so that better informed voters can put their knowledge to work — but thanks to the quorum, unknown bad candidates can't win by voting once for themselves or anything silly like that.)

(Oh, and speaking of unknowns, RV works well with write-in candidates, thanks to the quorum. Condorcet does not, since write-ins are implicitly in last place on every ballot by default.)

Re #2: Yes, there's some ga

Ranked Pairs is far and away the best option. It satisfies the monotonocity criterion. It is very simple to understand, especially from the perspective of a voter. You simply rank who you want to vote for. (Any voter who doesn't like it can simply rank their candidate #1 and leave the rest blank, just as we do now.)

I was a huge fan of Schulze method (another Condorcet variant, and one used in the real world in Debian elections), until RangeVoting.org convinced me today that Favorite Betrayal is an actual problem in any Condorcet method. Not only does Range Voting not have this problem, it also bypasses a lot of the Arrow's Impossibility Theorem baggage (since Arrow only applies to ranked systems, not scored systems) and, even better, is nearly as simple as Plurality or Approval.

(As the RV.org site points out, when you sit down and write code to implement a Condorcet method, any of them, you end up with about twice as much code compared to implementing RV. I've implemented Schulze voting before, so I know that of which they speak. Meanwhile, with RV, you just sum the votes as you would with Plurality or Approval, then divide by the number of voters. The fact that each vote is a ranged score rather than a {0|1} doesn't really change how the code works. Maybe you add a simple quorum check on the final score sums, but that's 2 lines of Perl using sort and grep.)

Plus bee swarms use it to vote on new hive locations, so even though I was skeptical at first about the Bayesian regret metric, RV is a tried and tested voting method in the real world, and bees are considerably less intelligent than even the dumbest voter.

There are two main problems with range voting. One, it makes voting overly complicated (not to the Slashdot crowd, but for the average voter.) Two, it is subject to gaming. Voters will be likely to rate secondary candidates lower than how they really feel in order to increase their top candidate's odds.

Re #1: I'd argue that RV is actually simpler than any ranked ballot. Seven words: Hot or Not, Olympic scoring, IMDb, Netflix. Every Joe-on-the-Street already understands RV. It's one of the three simplest systems possible (after Approval and Plurality). Try out this Range Vote on the US Presidential race — unlike Condorcet, you don't have to stand around hemming and hawing about exactly which order you'd put them in, plus you can actually express "I have no opinion" by leaving a line blank. (And since they threw in all the kitchen sink candidates on that poll, you'll use that option a lot.)

(In Condorcet, your only option for unknown candidates is to lump them all together in one big tie in the middle... but you're still saying "These unknown candidates are definitely worse than any of the ones I like" and "These unknown candidates are definitely better than any of the ones I hate". One of them could be {your undiscovered favorite|the reincarnation of Hitler}, but you've just {hurt|helped} their campaign. RV lets you leave them blank, so that better informed voters can put their knowledge to work — but thanks to the quorum, unknown bad candidates can't win by voting once for themselves or anything silly like that.)

(Oh, and speaking of unknowns, RV works well with write-in candidates, thanks to the quorum. Condorcet does not, since write-ins are implicitly in last place on every ballot by default.)

Re #2: Yes, there's some ga

Ranked Pairs is far and away the best option. It satisfies the monotonocity criterion. It is very simple to understand, especially from the perspective of a voter. You simply rank who you want to vote for. (Any voter who doesn't like it can simply rank their candidate #1 and leave the rest blank, just as we do now.)

I was a huge fan of Schulze method (another Condorcet variant, and one used in the real world in Debian elections), until RangeVoting.org convinced me today that Favorite Betrayal is an actual problem in any Condorcet method. Not only does Range Voting not have this problem, it also bypasses a lot of the Arrow's Impossibility Theorem baggage (since Arrow only applies to ranked systems, not scored systems) and, even better, is nearly as simple as Plurality or Approval.

(As the RV.org site points out, when you sit down and write code to implement a Condorcet method, any of them, you end up with about twice as much code compared to implementing RV. I've implemented Schulze voting before, so I know that of which they speak. Meanwhile, with RV, you just sum the votes as you would with Plurality or Approval, then divide by the number of voters. The fact that each vote is a ranged score rather than a {0|1} doesn't really change how the code works. Maybe you add a simple quorum check on the final score sums, but that's 2 lines of Perl using sort and grep.)

Plus bee swarms use it to vote on new hive locations, so even though I was skeptical at first about the Bayesian regret metric, RV is a tried and tested voting method in the real world, and bees are considerably less intelligent than even the dumbest voter.

There are two main problems with range voting. One, it makes voting overly complicated (not to the Slashdot crowd, but for the average voter.) Two, it is subject to gaming. Voters will be likely to rate secondary candidates lower than how they really feel in order to increase their top candidate's odds.

Re #1: I'd argue that RV is actually simpler than any ranked ballot. Seven words: Hot or Not, Olympic scoring, IMDb, Netflix. Every Joe-on-the-Street already understands RV. It's one of the three simplest systems possible (after Approval and Plurality). Try out this Range Vote on the US Presidential race — unlike Condorcet, you don't have to stand around hemming and hawing about exactly which order you'd put them in, plus you can actually express "I have no opinion" by leaving a line blank. (And since they threw in all the kitchen sink candidates on that poll, you'll use that option a lot.)

(In Condorcet, your only option for unknown candidates is to lump them all together in one big tie in the middle... but you're still saying "These unknown candidates are definitely worse than any of the ones I like" and "These unknown candidates are definitely better than any of the ones I hate". One of them could be {your undiscovered favorite|the reincarnation of Hitler}, but you've just {hurt|helped} their campaign. RV lets you leave them blank, so that better informed voters can put their knowledge to work — but thanks to the quorum, unknown bad candidates can't win by voting once for themselves or anything silly like that.)

(Oh, and speaking of unknowns, RV works well with write-in candidates, thanks to the quorum. Condorcet does not, since write-ins are implicitly in last place on every ballot by default.)

Re #2: Yes, there's some ga

Ranked Pairs is far and away the best option. It satisfies the monotonocity criterion. It is very simple to understand, especially from the perspective of a voter. You simply rank who you want to vote for. (Any voter who doesn't like it can simply rank their candidate #1 and leave the rest blank, just as we do now.)

I was a huge fan of Schulze method (another Condorcet variant, and one used in the real world in Debian elections), until RangeVoting.org convinced me today that Favorite Betrayal is an actual problem in any Condorcet method. Not only does Range Voting not have this problem, it also bypasses a lot of the Arrow's Impossibility Theorem baggage (since Arrow only applies to ranked systems, not scored systems) and, even better, is nearly as simple as Plurality or Approval.

(As the RV.org site points out, when you sit down and write code to implement a Condorcet method, any of them, you end up with about twice as much code compared to implementing RV. I've implemented Schulze voting before, so I know that of which they speak. Meanwhile, with RV, you just sum the votes as you would with Plurality or Approval, then divide by the number of voters. The fact that each vote is a ranged score rather than a {0|1} doesn't really change how the code works. Maybe you add a simple quorum check on the final score sums, but that's 2 lines of Perl using sort and grep.)

Plus bee swarms use it to vote on new hive locations, so even though I was skeptical at first about the Bayesian regret metric, RV is a tried and tested voting method in the real world, and bees are considerably less intelligent than even the dumbest voter.

There are two main problems with range voting. One, it makes voting overly complicated (not to the Slashdot crowd, but for the average voter.) Two, it is subject to gaming. Voters will be likely to rate secondary candidates lower than how they really feel in order to increase their top candidate's odds.

Re #1: I'd argue that RV is actually simpler than any ranked ballot. Seven words: Hot or Not, Olympic scoring, IMDb, Netflix. Every Joe-on-the-Street already understands RV. It's one of the three simplest systems possible (after Approval and Plurality). Try out this Range Vote on the US Presidential race — unlike Condorcet, you don't have to stand around hemming and hawing about exactly which order you'd put them in, plus you can actually express "I have no opinion" by leaving a line blank. (And since they threw in all the kitchen sink candidates on that poll, you'll use that option a lot.)

(In Condorcet, your only option for unknown candidates is to lump them all together in one big tie in the middle... but you're still saying "These unknown candidates are definitely worse than any of the ones I like" and "These unknown candidates are definitely better than any of the ones I hate". One of them could be {your undiscovered favorite|the reincarnation of Hitler}, but you've just {hurt|helped} their campaign. RV lets you leave them blank, so that better informed voters can put their knowledge to work — but thanks to the quorum, unknown bad candidates can't win by voting once for themselves or anything silly like that.)

(Oh, and speaking of unknowns, RV works well with write-in candidates, thanks to the quorum. Condorcet does not, since write-ins are implicitly in last place on every ballot by default.)

Re #2: Yes, there's some ga

Ranked Pairs is far and away the best option. It satisfies the monotonocity criterion. It is very simple to understand, especially from the perspective of a voter. You simply rank who you want to vote for. (Any voter who doesn't like it can simply rank their candidate #1 and leave the rest blank, just as we do now.)

I was a huge fan of Schulze method (another Condorcet variant, and one used in the real world in Debian elections), until RangeVoting.org convinced me today that Favorite Betrayal is an actual problem in any Condorcet method. Not only does Range Voting not have this problem, it also bypasses a lot of the Arrow's Impossibility Theorem baggage (since Arrow only applies to ranked systems, not scored systems) and, even better, is nearly as simple as Plurality or Approval.

(As the RV.org site points out, when you sit down and write code to implement a Condorcet method, any of them, you end up with about twice as much code compared to implementing RV. I've implemented Schulze voting before, so I know that of which they speak. Meanwhile, with RV, you just sum the votes as you would with Plurality or Approval, then divide by the number of voters. The fact that each vote is a ranged score rather than a {0|1} doesn't really change how the code works. Maybe you add a simple quorum check on the final score sums, but that's 2 lines of Perl using sort and grep.)

Plus bee swarms use it to vote on new hive locations, so even though I was skeptical at first about the Bayesian regret metric, RV is a tried and tested voting method in the real world, and bees are considerably less intelligent than even the dumbest voter.

There are two main problems with range voting. One, it makes voting overly complicated (not to the Slashdot crowd, but for the average voter.) Two, it is subject to gaming. Voters will be likely to rate secondary candidates lower than how they really feel in order to increase their top candidate's odds.

Re #1: I'd argue that RV is actually simpler than any ranked ballot. Seven words: Hot or Not, Olympic scoring, IMDb, Netflix. Every Joe-on-the-Street already understands RV. It's one of the three simplest systems possible (after Approval and Plurality). Try out this Range Vote on the US Presidential race — unlike Condorcet, you don't have to stand around hemming and hawing about exactly which order you'd put them in, plus you can actually express "I have no opinion" by leaving a line blank. (And since they threw in all the kitchen sink candidates on that poll, you'll use that option a lot.)

(In Condorcet, your only option for unknown candidates is to lump them all together in one big tie in the middle... but you're still saying "These unknown candidates are definitely worse than any of the ones I like" and "These unknown candidates are definitely better than any of the ones I hate". One of them could be {your undiscovered favorite|the reincarnation of Hitler}, but you've just {hurt|helped} their campaign. RV lets you leave them blank, so that better informed voters can put their knowledge to work — but thanks to the quorum, unknown bad candidates can't win by voting once for themselves or anything silly like that.)

(Oh, and speaking of unknowns, RV works well with write-in candidates, thanks to the quorum. Condorcet does not, since write-ins are implicitly in last place on every ballot by default.)

Re #2: Yes, there's some ga

Ranked Pairs is far and away the best option. It satisfies the monotonocity criterion. It is very simple to understand, especially from the perspective of a voter. You simply rank who you want to vote for. (Any voter who doesn't like it can simply rank their candidate #1 and leave the rest blank, just as we do now.)

I was a huge fan of Schulze method (another Condorcet variant, and one used in the real world in Debian elections), until RangeVoting.org convinced me today that Favorite Betrayal is an actual problem in any Condorcet method. Not only does Range Voting not have this problem, it also bypasses a lot of the Arrow's Impossibility Theorem baggage (since Arrow only applies to ranked systems, not scored systems) and, even better, is nearly as simple as Plurality or Approval.

(As the RV.org site points out, when you sit down and write code to implement a Condorcet method, any of them, you end up with about twice as much code compared to implementing RV. I've implemented Schulze voting before, so I know that of which they speak. Meanwhile, with RV, you just sum the votes as you would with Plurality or Approval, then divide by the number of voters. The fact that each vote is a ranged score rather than a {0|1} doesn't really change how the code works. Maybe you add a simple quorum check on the final score sums, but that's 2 lines of Perl using sort and grep.)

Plus bee swarms use it to vote on new hive locations, so even though I was skeptical at first about the Bayesian regret metric, RV is a tried and tested voting method in the real world, and bees are considerably less intelligent than even the dumbest voter.

There are two main problems with range voting. One, it makes voting overly complicated (not to the Slashdot crowd, but for the average voter.) Two, it is subject to gaming. Voters will be likely to rate secondary candidates lower than how they really feel in order to increase their top candidate's odds.

Re #1: I'd argue that RV is actually simpler than any ranked ballot. Seven words: Hot or Not, Olympic scoring, IMDb, Netflix. Every Joe-on-the-Street already understands RV. It's one of the three simplest systems possible (after Approval and Plurality). Try out this Range Vote on the US Presidential race — unlike Condorcet, you don't have to stand around hemming and hawing about exactly which order you'd put them in, plus you can actually express "I have no opinion" by leaving a line blank. (And since they threw in all the kitchen sink candidates on that poll, you'll use that option a lot.)

(In Condorcet, your only option for unknown candidates is to lump them all together in one big tie in the middle... but you're still saying "These unknown candidates are definitely worse than any of the ones I like" and "These unknown candidates are definitely better than any of the ones I hate". One of them could be {your undiscovered favorite|the reincarnation of Hitler}, but you've just {hurt|helped} their campaign. RV lets you leave them blank, so that better informed voters can put their knowledge to work — but thanks to the quorum, unknown bad candidates can't win by voting once for themselves or anything silly like that.)

(Oh, and speaking of unknowns, RV works well with write-in candidates, thanks to the quorum. Condorcet does not, since write-ins are implicitly in last place on every ballot by default.)

Re #2: Yes, there's some ga

Ranked Pairs is far and away the best option. It satisfies the monotonocity criterion. It is very simple to understand, especially from the perspective of a voter. You simply rank who you want to vote for. (Any voter who doesn't like it can simply rank their candidate #1 and leave the rest blank, just as we do now.)

I was a huge fan of Schulze method (another Condorcet variant, and one used in the real world in Debian elections), until RangeVoting.org convinced me today that Favorite Betrayal is an actual problem in any Condorcet method. Not only does Range Voting not have this problem, it also bypasses a lot of the Arrow's Impossibility Theorem baggage (since Arrow only applies to ranked systems, not scored systems) and, even better, is nearly as simple as Plurality or Approval.

(As the RV.org site points out, when you sit down and write code to implement a Condorcet method, any of them, you end up with about twice as much code compared to implementing RV. I've implemented Schulze voting before, so I know that of which they speak. Meanwhile, with RV, you just sum the votes as you would with Plurality or Approval, then divide by the number of voters. The fact that each vote is a ranged score rather than a {0|1} doesn't really change how the code works. Maybe you add a simple quorum check on the final score sums, but that's 2 lines of Perl using sort and grep.)

Plus bee swarms use it to vote on new hive locations, so even though I was skeptical at first about the Bayesian regret metric, RV is a tried and tested voting method in the real world, and bees are considerably less intelligent than even the dumbest voter.

There are two main problems with range voting. One, it makes voting overly complicated (not to the Slashdot crowd, but for the average voter.) Two, it is subject to gaming. Voters will be likely to rate secondary candidates lower than how they really feel in order to increase their top candidate's odds.

Re #1: I'd argue that RV is actually simpler than any ranked ballot. Seven words: Hot or Not, Olympic scoring, IMDb, Netflix. Every Joe-on-the-Street already understands RV. It's one of the three simplest systems possible (after Approval and Plurality). Try out this Range Vote on the US Presidential race — unlike Condorcet, you don't have to stand around hemming and hawing about exactly which order you'd put them in, plus you can actually express "I have no opinion" by leaving a line blank. (And since they threw in all the kitchen sink candidates on that poll, you'll use that option a lot.)

(In Condorcet, your only option for unknown candidates is to lump them all together in one big tie in the middle... but you're still saying "These unknown candidates are definitely worse than any of the ones I like" and "These unknown candidates are definitely better than any of the ones I hate". One of them could be {your undiscovered favorite|the reincarnation of Hitler}, but you've just {hurt|helped} their campaign. RV lets you leave them blank, so that better informed voters can put their knowledge to work — but thanks to the quorum, unknown bad candidates can't win by voting once for themselves or anything silly like that.)

(Oh, and speaking of unknowns, RV works well with write-in candidates, thanks to the quorum. Condorcet does not, since write-ins are implicitly in last place on every ballot by default.)

Re #2: Yes, there's some ga

Australia has had compulsory instant runoff voting (aka IRV, though we call it "preferential voting") for decades. It works pretty well. Systems like the Condorcet Method, Meek's Algorithm and Range Voting have some theoretical advantages, but they fail in one crucial respect: they are hard to count. Range Voting creates possibly hundreds of rounds of counting. [...]

Range Voting involves adding up a bunch of scores, then dividing by the number of votes cast to take the average. It's notionally simpler than IRV, only slightly more complicated than Plurality or Approval (there's one extra division at the end of the summation), it can use simpler voting machines than any ranked ballot system, plus the counting can be parallelized to individual districts (unlike IRV, in which all vote data must be physically located in one central location).

Oh, and real-world IRV examples (like Australia) have clearly shown that IRV still creates a two-party system, because it's so heavily impacted by strategic voting that third parties are still shut out. If too many people vote honestly when their #1 preference is a third party, they risk the elimination of their established-party fallback and throw the election to whichever established party they hate the most. That's the exact "split vote" phenomenon that switching away from Plurality is supposed to solve.

Whatever the solution is to two-party dominance, Australia has conclusively shown that IRV is not it.

(RangeVoting.org makes a credible argument that honeybees have already solved the problem, and the answer is Range Voting. For good or bad, I say this as someone who was for years 100% a Condorcet fan until today. I'll leave you to draw your own conclusions, but I do suggest you at least read the opposing arguments like I did.)

Australia has had compulsory instant runoff voting (aka IRV, though we call it "preferential voting") for decades. It works pretty well. Systems like the Condorcet Method, Meek's Algorithm and Range Voting have some theoretical advantages, but they fail in one crucial respect: they are hard to count. Range Voting creates possibly hundreds of rounds of counting. [...]

Range Voting involves adding up a bunch of scores, then dividing by the number of votes cast to take the average. It's notionally simpler than IRV, only slightly more complicated than Plurality or Approval (there's one extra division at the end of the summation), it can use simpler voting machines than any ranked ballot system, plus the counting can be parallelized to individual districts (unlike IRV, in which all vote data must be physically located in one central location).

Oh, and real-world IRV examples (like Australia) have clearly shown that IRV still creates a two-party system, because it's so heavily impacted by strategic voting that third parties are still shut out. If too many people vote honestly when their #1 preference is a third party, they risk the elimination of their established-party fallback and throw the election to whichever established party they hate the most. That's the exact "split vote" phenomenon that switching away from Plurality is supposed to solve.

Whatever the solution is to two-party dominance, Australia has conclusively shown that IRV is not it.

(RangeVoting.org makes a credible argument that honeybees have already solved the problem, and the answer is Range Voting. For good or bad, I say this as someone who was for years 100% a Condorcet fan until today. I'll leave you to draw your own conclusions, but I do suggest you at least read the opposing arguments like I did.)

Australia has had compulsory instant runoff voting (aka IRV, though we call it "preferential voting") for decades. It works pretty well. Systems like the Condorcet Method, Meek's Algorithm and Range Voting have some theoretical advantages, but they fail in one crucial respect: they are hard to count. Range Voting creates possibly hundreds of rounds of counting. [...]

Range Voting involves adding up a bunch of scores, then dividing by the number of votes cast to take the average. It's notionally simpler than IRV, only slightly more complicated than Plurality or Approval (there's one extra division at the end of the summation), it can use simpler voting machines than any ranked ballot system, plus the counting can be parallelized to individual districts (unlike IRV, in which all vote data must be physically located in one central location).

Oh, and real-world IRV examples (like Australia) have clearly shown that IRV still creates a two-party system, because it's so heavily impacted by strategic voting that third parties are still shut out. If too many people vote honestly when their #1 preference is a third party, they risk the elimination of their established-party fallback and throw the election to whichever established party they hate the most. That's the exact "split vote" phenomenon that switching away from Plurality is supposed to solve.

Whatever the solution is to two-party dominance, Australia has conclusively shown that IRV is not it.

(RangeVoting.org makes a credible argument that honeybees have already solved the problem, and the answer is Range Voting. For good or bad, I say this as someone who was for years 100% a Condorcet fan until today. I'll leave you to draw your own conclusions, but I do suggest you at least read the opposing arguments like I did.)

Also, in this scenario, in fact in ANY approval voting scenario, nobody ever has an incentive to disapprove their second choice while approving their third choice.

As shown on rangevoting.org, this statement is false in some scenarios. The linked page presents an example (see "Theorem 4") where there are four candidates (in rapidly decreasing popularity: A, B, C, D) and a voter has preferences C>>D>B>A, but the optimum strategy is to approve of C and B while dishonestly disapproving of D, because approving of D hurts your chances of breaking a tie in favor of C.

While you're never hurt by voting for your #1 favorite in Approval, there are weaker forms of strategic dishonesty lurking in the corners.

(reposted comment with correct formatting)

>Might as well just go with the simpler Approval voting... It's simpler, and
>more effective in my experience.

I partially agree. The most effective strategy under Range voting is to
always vote max or min score for each candidate that you think is a real
contender to win. Any other vote could be considered a partial abstention.
If the voting instructions are poor or minimal many voters will accidentally
partially abstain which will understandably make them angry. But if the instructions
are well written then I do not think that this will happen to a significant degree.

I like Approval voting but I see allowing partial abstentions as being a small improvement. I don't like the idea of encouraging frequent accidental partial abstentions so my support for Range Voting is very sensitive to the context and voting instructions.

Some more of my thoughts on this:
  http://allaboutvoting.com/2008/01/07/our-voting-system-is-a-loser/

See also the Range Voting advocacy site's comparison of Range vs. Approval and make up your own mind:
  http://www.rangevoting.org/rangeVapp.html

>Might as well just go with the simpler Approval voting... It's simpler, and >more effective in my experience. I partially agree. The most effective strategy under Range voting is to always vote max or min score for each candidate that you think is a real contender to win. Any other vote could be considered a partial abstention. If the voting instructions are poor or minimal many voters will accidentally partially abstain which will understandably make them angry. But if the instructions are well written then I do not think that this will happen to a significant degree. I like Approval voting but I see allowing partial abstentions as being a small improvement. I don't like the idea of encouraging frequent accidental partial abstentions so my support for Range Voting is very sensitive to the context and voting instructions. Some more of my thoughts on this: http://allaboutvoting.com/2008/01/07/our-voting-system-is-a-loser/ See also the Range Voting advocacy site's comparison of Range vs. Approval and make up your own mind: http://www.rangevoting.org/rangeVapp.html

Actually, that's wrong: Suppose you like Ron Paul best, Barack Obama almost as much, and Hillary least. If you were to vote Paul (100), Obama (100), Hillary (0), and the election was very close between Paul and Obama with Obama slightly ahead, your "strategic" voting would elect the candidate that you like slightly less.

Of course situations can happen where strategic voting pays off, but the simulations done by Warren D Smith involving mixtures of strategic and honest voters indicate that RV produces the best results (beating aproval, all studied condorset methods, plurality, etc.) of any studied system.

To be perfectly nit-picky: RV with instant runoff between the top two produces marginally better results than pure RV. Here's the comparison chart.

The reason range voting evades Arrow's Impossibility Theorem is discussed here: http://RangeVoting.org/ArrowThm.html

Range voting is quite common in questionnaires, where the form is often:

Q) You boss is an idiot.

[ ] Totally agree [ ] Partially agree [ ] Indifferent [ ] Partially disagree [ ] Totally diagree

I always answer those using the extremes for those cases where I'm not indifferent, in order to maximize the influence of my vote.

The range voting advocacy center acknowledge this as the optimal strategy in the generic case, but are able to find some corner cases where an honest voting strategy is better.

It is worth noting that Kuro5hin experienced the same effect, and switched from range voting to approval voting on comments.

For general elections, I'd recommend either approval voting (because the mechanics is so much simpler) or preferential voting because several of the vote counting techniques for preferential voting makes strategic voting very difficult.

• that they should be discussed and seriously considered
• that additional funding for basic research of E2E verifiable technologies should be supported
• that limited deployment of E2E verifiable systems should be allowed and encouraged
• laws that frustrate any of the above are misguided

• that they should be discussed and seriously considered
• that additional funding for basic research of E2E verifiable technologies should be supported
• that limited deployment of E2E verifiable systems should be allowed and encouraged
• laws that frustrate any of the above are misguided