Yes/No Voting vs. "classical" Approval Voting

[Bruce Gilson]

I use the term  "'classical' Approval Voting" to mean the system where, if you approve a candidate, you vote for him, while if you do not approve him, you do nothing in the box next to his name. It could be described as "If you're not for me, you're against me." And I've said numerous times that I don't like it because it offers no option to the voter who wants to indicate that a candidate is neither really good nor really bad. I've always preferred a system that allows a voter to abstain from the approve/disapprove count for such candidates - sometimes called "Yes/No Voting" because it requires a voter to explicitly put a "yes" or a "no" mark if he wants to be counted as approving or disapproving a candidate. (Since the term "Approval Voting" sometimes is used to include Yes/No Voting, as [for example] in Wikipedia, I need a term to describe the version I have termed "classical," which is why I've adopted that term.)

I think that besides the fact that Yes/No Voting provides an option for indicating that a candidate is neither really good nor really bad, which I think is absolutely necessary to correctly gauge the voters' opinions of the candidates, it may actually be easier to sell than classical AV because you often see people objecting to the latter because "it violates the one-man, one-vote rule." YNV is also something that can be easily set up on any voting machine that can handle referendum questions, which I am certain means any voting machine that is in use, so it is as easy to implement as classical AV, if not easier (a point often used in arguments against Score Voting).

[William Waugh]

This debate comes up over and over again. Do you want to allow write-ins? Would you give each voter the option of specifying that for write-ins that other voters write in but that that voter doesn't write in, their ballot should count with respect to that candidate as a vote against? Otherwise, there is the risk of electing a fanatic that hardly anyone knows about but who has 100 supporters.

On Saturday, February 6, 2016 at 9:20:44 AM UTC-5, Bruce R. Gilson started https://groups.google.com/forum/?fromgroups#!topic/electionscience/72PvkMyA008

[William Waugh]

It is not possible to design a voting system that correctly gauges the voters' opinions of the candidates.

[William Waugh]

In Approval Voting, a good strategy for handling your compromise candidate, I think, is to use a source of random numbers to construct a variable that has a 1% probability of evaluating to 0 and a 99% probability of evaluating to 1. Approve your compromise candidate if the value is 1, and otherwise, vote against your compromise candidate.

[William Waugh]

I don't know why you mentioned the "one person, one vote" principle in connection with your statement that you want to average over the "yes" or "no" votes and leave abstentions with respect to a given candidate out of the averaging for that candidate. In any event, in responding to a contention that Approval Voting violates "one person, one vote", I suggest to point out that the valuable core of the principle is that each person should be accorded equal political power to each other person with respect to the decision of who wins the office that is up for election (I'm only talking single-winner cases here), and that vote-for-one plurality violates that principle because it accords more power to the voter who is for one candidate and against the others than to a voter who is in favor of all the candidates but one. It permits the former voter to vote her exact opinion while preventing the latter voter from voting his.

[Bruce Gilson]

On Sat, Feb 6, 2016 at 10:07 AM, William Waugh <2knuw...@snkmail.com> wrote:

I don't know why you mentioned the

​​

"one person, one vote" principle in connection with your statement that you want to average over the "yes" or "no" votes and leave abstentions with respect to a given candidate out of the averaging for that candidate.

​I mentioned it because (if you read past posts) this issue has been raised by opponents of Approval Voting, not because I believe it does violate ​

​

"one person, one vote

​.​

"

[Bruce Gilson]

On Sat, Feb 6, 2016 at 10:01 AM, William Waugh <2knuw...@snkmail.com> wrote:

In Approval Voting, a good strategy for handling your compromise candidate, I think, is to use a source of random numbers to construct a variable that has a 1% probability of evaluating to 0 and a 99% probability of evaluating to 1. Approve your compromise candidate if the value is 1, and otherwise, vote against your compromise candidate.

​I see these "random choice" proposals over and over again, and I think they are the stupidest idea anyone has come up with in all the posts I've seen on this board. My vote should depend only on what I think of the candidates (whether their character, their chances of winning, their policies, or whatever), not on some random act like a spinner, a coin flip, etc.​

 

​The argument that, if a lot of voters all want to give someone .3 of a vote, 30% should give them 1 and 70% should give them 0 looks perhaps reasonable, but it ignores the fact that if I flip a fair coin 1000 times, the chance that there will be exactly 500 heads and 500 tails is really pretty small. (And corresponding things for any other randomizer. Throw a die 6 million times, and what are the odds of getting exactly 1 million 3's?) If anything, the only way to do something like this would be eliminate secrecy of the ballot, get all the people who want to give a candidate .3 of a vote, divide them into groups of 10 and say the first 3 of you vote yes, the remaining 7 vote no.

Sorry. A randomizer is NOT the solution to the approval voting dilemma.

[Bruce Gilson]

On Sat, Feb 6, 2016 at 9:53 AM, William Waugh <2knuw...@snkmail.com> wrote:

This debate comes up over and over again. Do you want to allow write-ins? Would you give each voter the option of specifying that for write-ins that other voters write in but that that voter doesn't write in, their ballot should count with respect to that candidate as a vote against? Otherwise, there is t

​​

he risk of electing a fanatic that hardly anyone knows about but who has 100 supporters.

"​T

​

he risk of electing a fanatic that hardly anyone knows about but who has 100 supporters" ​in this scheme is small compared to the risk, in classical AV, of defeating a candidate that might be a good compromise candidate but hasn't gotten his message across. All systems have their risks, but I think that the inability to distinguish between a candidate I sincerely oppose and one I don't know about is a serious flaw in classical AV, one that means I cannot support it.

 

[Bruce Gilson]

On Sat, Feb 6, 2016 at 9:56 AM, William Waugh <2knuw...@snkmail.com> wrote:

It is not possible to design a voting system that correctly gauges the voters' opinions of the candidates.

 

​In that case, why are we discussing voting systems at all​? Just randomly pick someone, as they did in some Greek city-states, to fill each office.

[William Waugh]

"...  but it ignores the fact that if I flip a fair coin 1000 times, the chance that there will be exactly 500 heads and 500 tails is really pretty small." -- that does not matter. The stats in a large election will work so the effect will be the same as would have happened with fine-grained Range Voting. The random technique allows the voter to exercise more political power than would be otherwise possible, and given Approval or otherwise overly coarse grain (e. g. five evenly spaced values), it provides the only strategy to evict the two-party system and elect a third party or an independent.

[William Waugh]

Someone who knows a candidate should not have more political power to support that candidacy than someone who does not know the candidate has to oppose that candidacy. You want to deny me the power to oppose the candidates I don't know. It is my civil right to oppose them in the election. I would compromise with you to the extent to allow you the freedom, on your ballot, to specify that you don't oppose candidates you don't know, and to have that choice honored in the tally. Will you compromise with me to allow me the corresponding freedom from the opposite viewpoint to specify that I oppose those candidates, if that's how I choose to exercise my political power, and have that choice honored in the tally as well? 

[William Waugh]

Because in Bayesian Regret studies, Approval does better than random choice.

The purpose of elections is to accord the citizens political power. 

[Bruce Gilson]

​Yes, but you seem to think that someone who does not know ​anything about a candidate is automatically disposed to oppose that candidate. I for one do not feel that way. If I do not know anything about a candidate, I do not

 

​want to vote against him -- but I do not want to vote for him either. (In fact, I have actually voted for a candidate I knew nothing about, on at least one occasion, but this was because the only other candidates on the ballot were people I specifically wanted to vote against.)

You say:

"

​

I would compromise with you to the extent to allow you

​​

the freedom, on your ballot, to specify that you don't oppose candidates you don't know, and to have that choice honored in the tally

,

"

 and ask, "

Will you compromise with me to allow me the corresponding freedom from the opposite viewpoint to specify that I oppose those candidates, if that's how I choose to exercise my political power, and have that choice honored in the tally as well?" In the system that I advocate

​, I do not see how it denies you freedom to vote against anyone you choose to vote against, even if it is because you don't know him. ​

​

​Whereas, in "classical" Approval V​oting, it does deny me "

​

the freedom, on

​ [my]

 ballot, to specify that

​[I]

 don't oppose candidates

​[I]

[Bruce Gilson]

​

You totally lose me here. What is the difference between ​"

​

gaug

​[ing]

 the voters' opinions of the candidates

​"​

 

​and ​"

​

accord

​[ing]​

the citizens political power

​"​? The only way I exercise any power in an election is to choose candidates who, in my opinion, are likely to act in the way I would desire.

[Steve Cobb]

The two-column ballot is more resistant to vote tampering, as a tamperer can only overlay votes if the voter leaves something blank.

Interesting that it might be usable with existing machines and software, given the similarity to referendum questions. 

[Toby Pereira]

Could this be used for approval as well as score voting? http://scorevoting.net/BetterQuorum.html

[Eric Sanders]

Why would you need a quorum in Approval Voting? You're simply adding singular 'approval votes,' so most 'votes' wins (just like now).

[William Waugh]

The difference is that if you are gauging opinion, you ask for opinion without any consequence except that you are going to publish the resulting statistics. If you are according power, you build a linkage from the voter's action to public policy, and you let the voter know that that linkage is there and how it will work. A rational and interested voter will respond by thinking about what strategy has the highest expected value in terms of how that voter values the different outcomes in terms of policy. The semantics of the voter's chosen input to the linkage is solely that input's effect on the outcomes as expected and understood by the voter, no matter how much as you would like to model it as a direct expression of the voter's opinion. Yes, the voter's opinion feeds into the voter's decision, but it is not necessarily reflected in a straightforward way in the action of the voter upon the link. The voter's strategic evaluation and decision intervenes between the voter's opinion and her vote.

I contend that evaluating competing voting systems based on how accurately they reflect the voter's opinion should come secondary to evaluating them based on whether they accord the voters equal power.

When a voting system accords the voters unequal political power, one of the ways it can do that is by according more power to voters who support fewer candidacies than to voters who support more count of candidacies. For example, vote-for-one Plurality has this characteristic. Such systems amplify the bandwagon effect, leading to two-party dominance, unless I am missing something.

[William Waugh]

On Sunday, February 7, 2016 at 1:04:18 PM UTC-5, Toby Pereira wrote:

Could this be used for approval as well as score voting? http://scorevoting.net/BetterQuorum.html

Yes, it could, but what bugs me about that kind of proposal is the magic number in it. How do you justify the choice of the count of fake ballots to introduce?

[William Waugh]

The context of Toby Pereira's (suggestive) question was a discussion in which Bruce R. Gilson was advocating that the approval votes should not be totaled, but rather averaged, with an option not to be counted either for or against a candidate. This conversation starts at https://groups.google.com/forum/?fromgroups=#!topic/electionscience/72PvkMyA008

[William Waugh]

... In the system that I advocate

​, I do not see how it denies you freedom to vote against anyone you choose to vote against, even if it is because you don't know him. ​

​

​...

The problem, if averaging is in effect and I as a voter don't have the option on my ballot to set what default applies to my ballot with respect to candidates not explicitly indicated thereon, is that I might not know that some particular fascist is running as a write-in, and so won't have the necessary knowledge to write in her name and mark it as "oppose".

[Bruce Gilson]

On Wed, Mar 2, 2016 at 12:03 AM, William Waugh <2knuw...@snkmail.com> wrote:

On Saturday, February 6, 2016 at 6:20:51 PM UTC-5, Bruce R. Gilson wrote:

On Sat, Feb 6, 2016 at 12:04 PM, William Waugh <2knuw...@snkmail.com> wrote:

On Saturday, February 6, 2016 at 11:20:04 AM UTC-5, Bruce R. Gilson wrote:

On Sat, Feb 6, 2016 at 9:56 AM, William Waugh <2knuw...@snkmail.com> wrote:

​​

It is not possible to design a voting system that correctly

​​

gauges the voters' opinions of the candidates.

 

​In that case, why are we discussing voting systems at all​? Just randomly pick someone, as they did in some Greek city-states, to fill each office.

Because in Bayesian Regret studies, Approval does better than random choice.

The purpose of elections is to

​​

accord the citizens political power. 

​

You totally lose me here. What is the difference between ​"

​

gaug

​[ing]

 the voters' opinions of the candidates

​"​

 

​and ​"

​

accord

​[ing]​

the citizens political power

​"​? The only way I exercise any power in an election is to choose candidates who, in my opinion, are likely to act in the way I would desire.

The difference is that if you are gauging opinion, you ask for opinion without any consequence except that you are going to publish the resulting statistics. If you are according power, you build a linkage from the voter's action to public policy, and you let the voter know that that linkage is there and how it will work. A rational and interested voter will respond by thinking about what strategy has the highest expected value in terms of how that voter values the different outcomes in terms of policy. The semantics of the voter's chosen input to the linkage is solely that input's effect on the outcomes as expected and understood by the voter, no matter how much as you would like to model it as a direct expression of the voter's opinion. Yes, the voter's opinion feeds into the voter's decision, but it is not necessarily reflected in a straightforward way in the action of the voter upon the link. The voter's strategic evaluation and decision intervenes between the voter's opinion and her vote.

​I still don't understand you, I guess. To me, "according power" as you describe it is simply ​making the opinion have its consequences, and so "gauging opinion" is a part of "according power." So you can't have "according power" without "gauging opinion," and your original statement that

"

​

It is not possible to design a voting system that correctly 

​​

gauges the voters' opinions of the candidates

​"​

 

​still means we have no reason to discuss voting systems at all.​

I contend that evaluating competing voting systems based on how accurately they reflect the voter's opinion should come secondary to evaluating them based on whether they accord the voters equal power.

​Now you've added the word "equal" here, which never got into the earlier discussion. And this really brings about a totally different question.

[Bruce Gilson]

​I see. I was not even thinking of write-in candidates. (In my state​

 

​I cannot vote for a write-in candidate, no matter how much I want to, unless he has formally declared as a write-in​ candidate, and there is a publicly available list of all such candidates, if any exist.) I would not object to adding a rule that all candidates whose names are not printed on the ballot are opposed unless explicitly written in. Everything I've said earlier was meant to apply to elections that don't have write-ins at all.

[Toby Pereira]

Just because of what was said in the opening post, about having to be for or against a candidate. This way an abstention is possible. If you allow good/bad/neutral, then it's beginning to turn into score voting (essentially a score out of 2).

[Toby Pereira]

It's not something I've really studied in great detail, but it certainly has a feel of "arbitrariness" about it. This obviously applies for score voting as well as approval. But this arbitrariness may be unavoidable if you want to allow abstentions. Other people on here would be in a better position to justify it. 

[William Waugh]

It's not something I've really studied in great detail, but it certainly has a feel of "arbitrariness" about it. This obviously applies for score voting as well as approval. But this arbitrariness may be unavoidable if you want to allow abstentions. Other people on here would be in a better position to justify it..

 "But this arbitrariness may be unavoidable if you want to allow abstentions."

Again, I suggest that a way to avoid any bad effects from magic numbers related to abstention (as well as to avoid all other problems related to abstention) is to make abstention optional for a voter, including in the case where the voter does not have a complete list of the candidates running, or make sure such a list is available. Either of such measures suffices to empower a voter who wants to oppose fully, every candidate whom the voter hasn't studied enough to convince herself the candidate is not unacceptably horrible according to the voter's values.

[William Waugh]

On Wed, Mar 2, 2016 at 9:06 AM, Bruce Gilson brgster-at-gmail.com |google electoral public/pol| <9fh9w...@sneakemail.com> wrote:

Great. You and I are now in accord on abstention.