Challenge to Score voting advocates

[scott...@gmail.com]

I’ve read some of the information on Scored voting. I understand it. In all candor, I’m not impressed.  I do not see the superiority of Scored voting, especially given a credible alternative--Reciprocal voting.

Score voting depends on the honesty of the voters. Once the parties figure out how to game it the elective system will be dominated by gaming practices.  Specifically, voters of one party will be asked to give their candidates high scores, and candidates of the alternate parties low or zero scores. Honest, less partisan voters, will swing the election on way or the other.  Here in the US at least, this will eventually become little different than the current plurality system.  

In contrast, in Reciprocal voting voters give one “for” vote for one candidate, and one “against” vote for one candidate. Reciprocal voting has some interesting elective properties, the most important being that it is difficult to game.  It all but eliminates the two party system and better attaches the loyalty of office holders to the represented region. Reciprocal is effective over the long run where the efficacy of Score is rather limited due to the predictable gaming that will occur.

With Reciprocal voting, loyalist party members will give their “against” vote to the dominant opposing candidate.In a two party system this causes the members of the  parties to negate each other’s dominant candidate.  Either a second best party candidate, or a third party candidate will win.  Over time the only way candidates can be elected is if they bring strong positive utility to the table, and do not have strong negatives against them.  Negatives will include things that interfere with regional loyalty such as excess party loyalty or excessive obligations incurred by receiving campaign donations.  

Over time candidates elected under Reciprocal will truly represent the people of their region.  The representatives become people who are respectable, loyal to the people of their district, and likeable. Duverger’s law may be negated. Parties will become very similar if not nonexistent and third parties may become viable.

Gaming is made difficult in Reciprocal. When it becomes apparent that a candidate is being gamed into office the candidate becomes a target for receiving “against” votes from the non-supporting voters. Any time propaganda becomes obvious and it is intended to favor one candidate, that candidate gets a lot of “against” votes.  In this system guilt by association leads to vote loss, so campaigning changes from being mean and accusatory to being nice and promotional.  It is the tendency to vote against candidates who appear less than scrupulous, less than faithful to the voters or simply mean spirited that gives Reciprocal voting its reciprocal character.

If a negative vote were added to any voting system where the voters may vote for all candidates such as Approval or Score, the end results would not change.  An Approval vote system with an “against” vote added to it would not produce better results as Reciprocal.  Neither would a Score voting system that was changed to a +/- scale.  Those systems would still be just as gameable and still approximate Plurality eventually.

I have to challenge any proposal that Score voting would improve much if anything in the character of elected office holders. I have to challenge the proposal that office holders would more accurately represent the will of the people. I can only believe that Score voting would just continue the existing system, but with a fancier looking ballot and a more expensive elective process.  

Reciprocal voting is a credible system for creating more moderate, less divisive, more faithful, more competent representatives.  

If you can come up with a credible way that Reciprocal could be gamed (given that gaming, when discovered, leads to “against” votes), please let us know.

If you can explain how Score voting will not be gamed so it becomes a variation on Plurality, let us know, 'cuz I'm not seeing it. You have to assume that most voters will not vote strategically, when in actuality most voters will.

Enjoy!

[Clay Shentrup]

On Monday, June 27, 2016 at 12:41:41 PM UTC-7, scott...@gmail.com wrote: 

I do not see the superiority of Scored voting, especially given a credible alternative--Reciprocal voting.

These Bayesian Regret figures show the massive superiority of Score Voting vs. what you're calling "reciprocal voting" (called "VtForAgainst" in the green table).

http://scorevoting.net/StratHonMix.html

Score voting depends on the honesty of the voters.

So does every deterministic voting method. But as those Bayesian Regrets show, Score Voting behaves better with 50% strategic voting than your proposed system does with 100% honest voting.

Range 0.16329 0.04802

VtForAgainst 0.41726 0.20101

 

Once the parties figure out how to game it the elective system will be dominated by gaming practices.

Yeah, we've heard these tropes for the past decade+, and they've been debunked over and over again.

Parties don't vote. Voters do. And they're much more likely to vote in their own self interest than to do whatever's best for their own party. Just look at Ralph Nader. He told his supporters to vote for him, but most of them voted for Gore. Similar things will happen with Score Voting.

Specifically, voters of one party will be asked to give their candidates high scores, and candidates of the alternate parties low or zero scores.

Of course they will. But they won't do it unless it's in their own self interest, which it usually won't be.

 

Honest, less partisan voters, will swing the election on way or the other.  Here in the US at least, this will eventually become little different than the current plurality system.

With Plurality Voting, strategic voters are those who do not vote for their favorite candidate.

You're saying that with Score Voting, strategic voters will only vote for their favorite candidate.

And, somehow, you don't see the contradiction when you say Score Voting will, because of tactical voters, become little different than the current Plurality system.

You are committing the same naive fallacy that every newcomer to voting methods commits. Please think about this a little deeper before spending more time trying to debate experts about it.

[Toby Pereira]

Have you done much investigation into how people would vote under reciprocal voting, and what the results would be, or is this intuition?

You ask why score voting won't be gamed so that it becomes a version of plurality - many people would just give minimum and maximum scores (approval style), and some of these would give the maximum to just one candidate (so plurality style), but that's not the end of the world. It's never strategically disadvantageous to give a maximum score to your favourite candidate, so those who would vote strategically vote for a frontrunner who isn't their true favourite under plurality, could give a maximum score to their favourite of the frontrunners as well as their true favourite. If there is a brilliant candidate who people might be put off voting for under plurality, they would have no reason not to give them a top score under score voting.

[Kevin Baas]

range voting fails later-no-harm so it's an ineffective measure against gerrymandering.  having said that, what you're proposing is just a special case of range voting.  it's range voting with max 1 positive vote and max 1 negative point.  only reason it's not later-no-harms because there is no later.  still, neither surplus nor wasted votes are transferred, so it results in majority rule.  so it's a non-solution to gerrymandering.  it makes multi seat districts worse than single seat.  

both systems lake vote transfers so imo they're no use for multi-member districts

[Clay Shentrup]

On Monday, June 27, 2016 at 4:13:00 PM UTC-7, Kevin Baas wrote:

range voting fails later-no-harm so it's an ineffective measure against gerrymandering.

Later-no-harm is more of a flaw than a feature. It requires a voting method to ignore important data.

I don't know how you think it has any bearing on Gerrymandering though. To fight Gerrymandering with a voting method, you have to have proportional representation.

[Kevin Baas]

yeah i was struggling to word that - there's a correlation to later-no-harm and non-proportional.  really i'd say the more direct flaw is lack of transferable votes.

[William Waugh]

In a second thread, https://groups.google.com/d/msg/electionscience/7B5aLd33Yr0/FM58dV2DBgAJ , scott97007-at-gmail.com again refers to a system he wants to propose, Reciprocal Voting.

I understand that the form of the ballot is that the voter can choose one candidate to support and another candidate to oppose. So, in all, each candidate's status with respect to a voter is either of Supported, Opposed, or unmentioned, so there are three possibilities. Given these possibilities, they are not independently chosen with respect to each candidate by the voter (unlike Score Voting), but there are constraints, that the voter can use her up-grade for only one candidate and her down-grade for only one candidate.

However, again, I don't see the tallying algorithm. What is it?

In regard to your critique of Score Voting (also known as Range Voting) that voters will not vote "honestly", I think it is incorrect to describe some strategies as "honest" and others as "lies", even though the greatly respected Warren D. Smith talks about votes that "lie". Whatever voting system is in place, we can generally expect voters to try to make the choice that seems to them likely to move the needle in the direction they prefer, in terms of the electoral outcome. In fact, I think Game Theory predicts voter behavior. To evaluate two voting systems against one another, we have to ask ourselves in each case what strategy would work to the voters's advantage, and try to approximate that strategy. Then assume the voters use those strategies, see what winners the systems produce given that, and calculate or simulate the Bayesian Regret on that basis.

Choose-one Plurality voting does not offer a strategy to defeat two "front-runners" anointed by money and other strong-arm factors.

I say that Score does offer such a strategy, which I will outline below (again) in brief.

I don't know whether Reciprocal Voting offers such a strategy, because I still don't know what Reciprocal Voting is. I need the tallying algorithm in order to evaluate it.

In the 2012 POTUS vote, about 1% of the voters voted for anyone who wasn't in the two big gangs. Assume we have Score Voting with therefore a granularity of 1%. A voter who wants to defeat the gangs has in mind one or more truly approved (by that voter) candidates, and also some preference regarding whether if one of the gang candidates win, which it should be, Hitler or Mussolini in effect. The vote-for-one system works for the gangs by placing the voter in a fix where voting for a preferred candidate instead of Mussolini results in a fear of electing Hitler, and indeed, other voters are likely to blame this voter (if she reveals and argues for her strategy) for contributing to Hitler's victory. Witness all the propaganda that said a vote for Nader was a vote for the Republicans. This is still used against the Greens today. But back to the Score Voting case. The voter has in mind a collection of truly preferred candidates and a compromise candidate, Mussolini. What strategy will lead to the desired result? Suppose the range is expressed (any y=mx+b mapping is equivalent) as {100, 98, 96, ..., -98, -100}. This satisfies the constraint of 1% granularity that I mentioned. I say the strategy is to score the preferred candidates 100, Mussolini 98, and all others -100. You say you expect the voters to bullet vote. I say they will not, because doing so would not be in their interest. Doing that would sacrifice their right to express a preference for Mussolini over Hitler. Or if they used the bullet vote another way, it would sacrifice their right to support their preferred candidates. Voting the way I recommend would result in the preferred candidates getting more support each election, so eventually those who agree for the defeat of the gangs can start to ease off on their level of support for Mussolini, so eventually a non-gang candidate wins. This is not possible with the choose-one system. It permits no gradualism, only a sudden break, as when the Whig party winked out of existence and the Republican party took its place. The two-gang system continued, just with different names.