Compromise between Approval Voting and Range Voting advocates
168 views
Skip to first unread message
parker friedland
May 14, 2017, 12:41:41 AM5/14/17
to The Center for Election Science
There is a rift between electoral reform advocates as to whether approval voting or range voting is actually better:
In Approval's Defense:
In range voting, voters who vote strategically often have their votes count for less. As a result, strategic voters are able to tip the scales more then the average voter. Many of the original advocates of plurality voting in US elections during the American revolution believed that the desire for many voter's to be honest would prevent plurality voting from devolving into polarizing elections between two ideologies were wrong. As such, it is likely that range voting could deteriorate into approval voting. Also even though many of us in the electoral science community prefer utility over majority, most normal people tend to think that having a majority of the votes is more fair then having a greater voting utility in a two person race. As such, even though I am a supporter of improving voter efficiency, I believe that approval voting is more resistant to "anti-majority catastrophe" propaganda because voters just don't vote for both or none in that scenario, which makes that kind of propaganda less effective.
In Range's Defense:
Because voter's are already so used to using plurality voting, If you present them with a ballot that looks exactly the same as an approval ballot, they will vote exactly the same as in approval voting. The institute of Electoral Engineering discovered that 80% percent of voter's were voting in their election the same way they would in an approval voting. Unfortunately as a result, they got rid of approval voting. Thus, range voting is needed in order to break voters out of their simple minded plurality voting ways.
The Compromise:
Net Approval voting (otherwise known as Favorability Voting) presents voters with a radically different ballot then they are used to in plurality voting. Each candidate has three circles next to their name. Inside each circle is either a check mark, a question mark, or an X. At the top of the ballot, the column of check circles are labeled with "approve". The next column of circles are labeled with "unsure or no opinion". The third column of circles are labeled with "disapprove". Each voter's favorability score consists of their total approval votes minus their disapproval votes divided by the entire population. Many range advocates might realize that this is the exact same thing as range voting with the scores 0, 1, and 2. However I believe in Net approval voting, it is important to use the scores -1, 0, and +1, in order to convince voters that have no opinion about a candidate that giving them a score in the middle of the range of possible scores for that candidate will have a neutral effect on that candidate (the same effect as not even showing up to vote would have for that candidate). This voting system will also be more resistant to "anti-majority catastrophe" propaganda because in a two person race, if nobody used the no opinion/unsure rating, then the election would result in a plurality like race. You can then convince majority fans that it's OK to let some people to purposefully let their ballot count for less if they don't know who a candidate is and don't want to affect their score. There is also similar form of approval disapproval voting that just compare's the ratio of approval votes to disapproval votes for a particular candidate however that kind of voting can result in a participation paradox which opens up an entirely new way of criticizing the election method. You may disapprove of my recommendation of Net Approval voting however all I am trying to do is find an approval/range based voting method that is the most likely to get the support it needs to be implemented in real life.
In Approval's Defense:
In range voting, voters who vote strategically often have their votes count for less. As a result, strategic voters are able to tip the scales more then the average voter. Many of the original advocates of plurality voting in US elections during the American revolution believed that the desire for many voter's to be honest would prevent plurality voting from devolving into polarizing elections between two ideologies were wrong. As such, it is likely that range voting could deteriorate into approval voting. Also even though many of us in the electoral science community prefer utility over majority, most normal people tend to think that having a majority of the votes is more fair then having a greater voting utility in a two person race. As such, even though I am a supporter of improving voter efficiency, I believe that approval voting is more resistant to "anti-majority catastrophe" propaganda because voters just don't vote for both or none in that scenario, which makes that kind of propaganda less effective.
In Range's Defense:
Because voter's are already so used to using plurality voting, If you present them with a ballot that looks exactly the same as an approval ballot, they will vote exactly the same as in approval voting. The institute of Electoral Engineering discovered that 80% percent of voter's were voting in their election the same way they would in an approval voting. Unfortunately as a result, they got rid of approval voting. Thus, range voting is needed in order to break voters out of their simple minded plurality voting ways.
The Compromise:
Net Approval voting (otherwise known as Favorability Voting) presents voters with a radically different ballot then they are used to in plurality voting. Each candidate has three circles next to their name. Inside each circle is either a check mark, a question mark, or an X. At the top of the ballot, the column of check circles are labeled with "approve". The next column of circles are labeled with "unsure or no opinion". The third column of circles are labeled with "disapprove". Each voter's favorability score consists of their total approval votes minus their disapproval votes divided by the entire population. Many range advocates might realize that this is the exact same thing as range voting with the scores 0, 1, and 2. However I believe in Net approval voting, it is important to use the scores -1, 0, and +1, in order to convince voters that have no opinion about a candidate that giving them a score in the middle of the range of possible scores for that candidate will have a neutral effect on that candidate (the same effect as not even showing up to vote would have for that candidate). This voting system will also be more resistant to "anti-majority catastrophe" propaganda because in a two person race, if nobody used the no opinion/unsure rating, then the election would result in a plurality like race. You can then convince majority fans that it's OK to let some people to purposefully let their ballot count for less if they don't know who a candidate is and don't want to affect their score. There is also similar form of approval disapproval voting that just compare's the ratio of approval votes to disapproval votes for a particular candidate however that kind of voting can result in a participation paradox which opens up an entirely new way of criticizing the election method. You may disapprove of my recommendation of Net Approval voting however all I am trying to do is find an approval/range based voting method that is the most likely to get the support it needs to be implemented in real life.
parker friedland
May 14, 2017, 1:02:09 AM5/14/17
to The Center for Election Science
I suggest that we should do a poll on how supportive Americans would be of Net Approval Voting so we can at least compare it with other electoral reforms
parker friedland
May 14, 2017, 2:47:49 PM5/14/17
to The Center for Election Science
There are other ways to attack Range Voting. I looked at sight-line's propaganda about Range Voting and one of their criticisms was that "Trying to calculate your honest scores for each candidate, and then also
understand the best strategy under Score Voting, would put an even higher cognitive burden on voters than Approval Voting." I believe that Score Runoff voting can also be attacked in this way. Net Approval Voting and Favorability Runoff Voting (Net Approval but with runoff) are more resistant to this attack because it is more straight forward that you do not want to give a candidate a score of no opinion unless you want to have less impact on the election.
David Hollander
May 15, 2017, 7:23:55 PM5/15/17
to The Center for Election Science
> then also
understand the best strategy under Score Voting, would put an even higher cognitive burden on voters than Approval Voting
Cognitive science indicates people have sufficient working memory to consider 7 +/- 2 independent alternatives simultaneously [Miller 1956]. This indicates Range-5 should work great. A decision between 5 scores should be within the working memory bounds for nearly the entire population without the use of chunking heuristics, and cannot be criticized as having too high of a cognitive burden. The cognitive burden for ranked voting is actually much higher than for range-5 voting if there is a large number of candidates on the ballot.
> However I believe in Net approval voting, it is important to use the scores -1, 0, and +1
I believe this scale is also referred to as "Evaluative Voting". For Range 5, the evaluative scale would be {-2,-1,0,+1,+2}. The expanded scale would still allow voters to determine an appropriate score using binary questions:
1) Do I support or oppose the alternative? Support --> Positive. Oppose --> Negative.
2) Is my view strongly or weakly substaniated? Strongly --> Two. Weakly --> One.
3) Was I able to answer these questions? Yes --> Non-zero. No --> Zero.
Range 5 would be more expressive in Range 3 and cannot be criticized as having too high of a cognitive burden. The numbers in a Range-5 signed integer scale could be explained to voters as intuitively corresponding to "strongly oppose", "oppose", "abstain", "support", "strongly support". The additional expressiveness makes it easier for voters to indicate partial positive support for a candidate. It avoids the strategic criticisms of approval voting that it is biased towards centrists who campaign on receiving "token" votes from voters who do not actually expect the candidate to win.
1) Do I support or oppose the alternative? Support --> Positive. Oppose --> Negative.
2) Is my view strongly or weakly substaniated? Strongly --> Two. Weakly --> One.
3) Was I able to answer these questions? Yes --> Non-zero. No --> Zero.
Range 5 would be more expressive in Range 3 and cannot be criticized as having too high of a cognitive burden. The numbers in a Range-5 signed integer scale could be explained to voters as intuitively corresponding to "strongly oppose", "oppose", "abstain", "support", "strongly support". The additional expressiveness makes it easier for voters to indicate partial positive support for a candidate. It avoids the strategic criticisms of approval voting that it is biased towards centrists who campaign on receiving "token" votes from voters who do not actually expect the candidate to win.
Warren D Smith
May 15, 2017, 7:36:01 PM5/15/17
to electio...@googlegroups.com
Actually, the "can remember 7+-2 things" alleged cognitive limit
heavily hurts attempts to poll people by telephone about more than 7
candidates if one uses any rank-order ballot system.
But it does NOT particularly hurt score-voting style telephone polls,
because you do not need to keep everybody else in mind as a pollee,
you simply keep the candidate the pollster is asking about right now,
in mind, and score him. Then repeat.
If IRV style voting, then you could not do that,
because your ranking for somebody cannot
be independent of your decisions for all the others.
Now think about approval voting. With approval,
your decision about where to place
your personal "dividing line" for acceptable vs not, DEPENDS on
who is in the race. It is a strategic decision.
I.e. even if you plan to approval-vote wholy honestly, you still need
to make a strategic decision which depends on knowing the full
slate of candidates. So ironically, this cognitive limit
could actually make approval voting be more difficult
(in, e.g. a telephone polling setting) than score voting!
---
By the way, a related cognitive limit was the "digit span" test where
you have to remember for a short time a sequence of N random digits.
What is your N value? I used to be a lot better than the average person
at this test with N=15 or so, but I think my brain is decaying now
and I may now be comparable or worse than the average person.
A fairly dramatic decline.
--
Warren D. Smith
http://RangeVoting.org <-- add your endorsement (by clicking
"endorse" as 1st step)
heavily hurts attempts to poll people by telephone about more than 7
candidates if one uses any rank-order ballot system.
But it does NOT particularly hurt score-voting style telephone polls,
because you do not need to keep everybody else in mind as a pollee,
you simply keep the candidate the pollster is asking about right now,
in mind, and score him. Then repeat.
If IRV style voting, then you could not do that,
because your ranking for somebody cannot
be independent of your decisions for all the others.
Now think about approval voting. With approval,
your decision about where to place
your personal "dividing line" for acceptable vs not, DEPENDS on
who is in the race. It is a strategic decision.
I.e. even if you plan to approval-vote wholy honestly, you still need
to make a strategic decision which depends on knowing the full
slate of candidates. So ironically, this cognitive limit
could actually make approval voting be more difficult
(in, e.g. a telephone polling setting) than score voting!
---
By the way, a related cognitive limit was the "digit span" test where
you have to remember for a short time a sequence of N random digits.
What is your N value? I used to be a lot better than the average person
at this test with N=15 or so, but I think my brain is decaying now
and I may now be comparable or worse than the average person.
A fairly dramatic decline.
--
Warren D. Smith
http://RangeVoting.org <-- add your endorsement (by clicking
"endorse" as 1st step)
Jameson Quinn
May 15, 2017, 7:57:08 PM5/15/17
to electionsciencefoundation
I'm conflicted about responding to this thread. I think that we voting reform activists spend too much time arguing between essentially similar systems, and not enough time planning how to actually get them implemented.
But. This is a thread about systems.
I agree with the premise. Approval and score are both good voting systems, but neither is perfect. Approval's lack of expressiveness is unsatisfying to some. Others (or sometimes the same people) are concerned that, under score, naive voters may unintentionally give up voting power. (And in fact, if strategic asymmetry is a potential problem in score, it's also probably a problem to some degree in approval, insofar as those same naive voters use approval probabilistically.)
So a compromise is called for.
But adding abstention, and/or using score with just 3 or 5 possible scores, doesn't really solve the problems.
I believe that the problems are solvable, with methods like score runoff voting¹ or 3-2-1 voting. These methods are more expressive than approval, and also make it easy for a voter to cast a vote that's strategically optimal in terms of deciding between the top two candidates (as defined by the earlier steps of the method). Both methods have good simulated Voter Satisfaction Efficiency, and are also resistant to strategy in simulations. With 3-2-1 voting, you also get an extremely simple ballot, so that there's almost no room for strategy; the robustness is excellent. With SRV, you get more expressivity and thus slightly better VSE for honest voters.
So from a theoretical point of view, it's still great to talk about different single-winner systems, and breakthroughs are still possible. In just the last few years, the "improved Condorcet" family of methods, and SODA, have shown property combinations that may have seemed impossible before. But in terms of practical reform, I think there are really 4 single-winner methods worth talking about. IRV, for its track record; Approval, for its simplicity; and SRV and 3-2-1, for their combination of good properties with reasonable simplicity.
(Personally, if it were up to me, I'd never choose IRV, unless plurality were the only other option. But however I feel about it, it deserves a place on that list.)
Jameson
¹It seems that SRV is now being called star voting; not to be confused with my earlier multiwinner proposal of STAR voting, which I've now tweaked slightly and am calling OL/D voting.
--
You received this message because you are subscribed to the Google Groups "The Center for Election Science" group.
To unsubscribe from this group and stop receiving emails from it, send an email to electionscience+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Mark Frohnmayer
May 15, 2017, 8:14:34 PM5/15/17
to electionsciencefoundation
Some personal observations after running an election with 0-5 Star (Score + Automatic Runoff) Voting aka Score Runoff aka SRV:
People get 0-5 scoring. Not even a question. From Amazon to every annoying app on your phone, 0-5 requires very little explanation. We started to see real pushback from lay voters and reform advocates when the scale was 0-9, saying it'd require too much thinking and "strategy." Nobody in the audience of ~35 had a lack of understanding of the 0-5 scale, and voters completed ballots in almost real time.
I agree regarding expressiveness: in petitioning in 2014, a real point of pushback I got on approval is that you can't express any kind of preference order between approved candidates. When it comes to VSE, 3-2-1 performs better with 3 options than SRV, SRV performs better than 3-2-1 with 4 or more score options.
Jameson, I don't think the folks pushing for the name change had come across your prior proposal. And as aside I have been meaning to check out OL/D, just looking for a little extra bandwidth :-).
To unsubscribe from this group and stop receiving emails from it, send an email to electionscien...@googlegroups.com.
--
You received this message because you are subscribed to the Google Groups "The Center for Election Science" group.
To unsubscribe from this group and stop receiving emails from it, send an email to electionscien...@googlegroups.com.
David Hollander
May 15, 2017, 8:29:55 PM5/15/17
to The Center for Election Science
On Monday, May 15, 2017 at 6:36:01 PM UTC-5, Warren D. Smith (CRV cofounder, http://RangeVoting.org) wrote:
> it does NOT particularly hurt score-voting style telephone polls, because you do not need to keep everybody else in mind as a pollee,
> it does NOT particularly hurt score-voting style telephone polls, because you do not need to keep everybody else in mind as a pollee,
It is true that the pollee does need to keep in mind each candidate when stating a score. However, they do need to keep in mind the range of possible scores. The number of candidates on the ballot and the range of possible scores
which may be assigned to each candidate are orthogonal considerations. With a large score range greater than Range-7, voters will be more likely to employ heuristics in order to determine the score they assign to each candidate. Heuristics such as the availability heuristic and anchoring-and-adjustment heuristic may introduce bias and decrease voter honesty. For an extremely large number range, political consultants will inform their candidate to verbally use the highest number on the scale as frequently as possible. This is in order to associate their candidate with this number in the minds of voters, so that voters use it as an anchor point when employing common numerical heuristics.
> With approval, your decision about where to place your personal "dividing line" for acceptable vs not, DEPENDS on who is in the race.
I agree. There is also a lower bound on the appropriate range. If the range is so small that it results in "lossy compression", then the voter is effectively dropping bits of information from their preference encoding. If they are forced to drop a large number of bits, they will want to consider how this affects the quality of the final results, in order minimize the perceptible loss of information.
> "digit span" test
Memorization and recall of a sequence may not be as relevant to score voting as identification of a unique value. There are many compression algorithms and strategies available for ordered sequences which would not be relevant for score voting. I believe one such strategy involves chunking subspans of digits into imaginary pictures which form a narrative. The number of chunking operations which voters should be required to perform to compress the possible range of numbers in range voting should perhaps be minimized, as chunking decisions may be performed using biased heuristics spread by political campaign consultants in order to distort results.
> "digit span" test
Memorization and recall of a sequence may not be as relevant to score voting as identification of a unique value. There are many compression algorithms and strategies available for ordered sequences which would not be relevant for score voting. I believe one such strategy involves chunking subspans of digits into imaginary pictures which form a narrative. The number of chunking operations which voters should be required to perform to compress the possible range of numbers in range voting should perhaps be minimized, as chunking decisions may be performed using biased heuristics spread by political campaign consultants in order to distort results.
parker friedland
May 15, 2017, 9:26:15 PM5/15/17
to The Center for Election Science
The more you expand the range, the most common ratings voters will assign to each candidate will become max, min, and 0 (assuming your using the evaluative scale). Heuristics do become a problem after that and after having at least 5 values to rate each candidate, the information gained as a result of adding more values becomes minuscule and it does still seem a little useful to provide voters with the values in between 0, max, and min. I believe net approval is the most reasonable because it cuts straight to the point by providing voters the scores that actually matter however a point can be made about token candidates that makes a 5 valued evaluative scale also pretty reasonable. I don't believe anything beyond that is reasonable.
parker friedland
May 15, 2017, 9:31:38 PM5/15/17
to The Center for Election Science
In fact huge corporations like google and Netflix seem to agree with me on that because their most common ranges are ranges of 5 (example: Google play store), 3 (Netflix, YouTube), and 2 (Facebook likes)
parker friedland
May 15, 2017, 9:41:47 PM5/15/17
to The Center for Election Science
However when it comes to voting, I strongly prefer the evaluative scale {-2,-1,0,+1,+2} / {-1, 0, 1} over the Google scale. In fact I believe that in ranges of 5 and 3, this scale will decrease use of the Max/Min voter heuristics because 0 is a much more meaningful reference point then 1 and 3 are.
parker friedland
May 15, 2017, 10:04:23 PM5/15/17
to The Center for Election Science
I put YouTube under a range of 3 (net approval voting) however right now YouTube's algorithm does not care about dislikes very much. In fact it slightly favors video's that have been disliked rather then not liked or disliked even when both are watched for the same amount of time:
(warning: skip to 0:10s because of this guy's really annoying intro)
https://www.youtube.com/watch?v=Om4c4ilT_6g
(warning: skip to 0:10s because of this guy's really annoying intro)
https://www.youtube.com/watch?v=Om4c4ilT_6g
David Hollander
May 15, 2017, 10:58:19 PM5/15/17
to The Center for Election Science
I have attached and linked a mockup of a Range-5 ballot using the evaluative \ signed-integer scale in case it is of interest:
https://i.imgur.com/Xbpub8s.png
https://i.imgur.com/Xbpub8s.png
parker friedland
May 16, 2017, 12:43:42 AM5/16/17
to The Center for Election Science
Because word doesn't always display documents correctly depending on what version of word you have and if you have enable editing on, here is the ballot in PDF form:
Hélène Mayeur
May 16, 2017, 5:41:38 AM5/16/17
to electio...@googlegroups.com
Looks like the ballot in the Vote de Valeur experiment in 2012. (http://www.votedevaleur.org/)
De : David Hollander <dhl...@gmail.com>
À : The Center for Election Science <electio...@googlegroups.com>
Envoyé le : Mardi 16 mai 2017 6h09
Objet : Re: [CES #16210] Re: Compromise between Approval Voting and Range Voting advocates
Hélène Mayeur
De : David Hollander <dhl...@gmail.com>
À : The Center for Election Science <electio...@googlegroups.com>
Envoyé le : Mardi 16 mai 2017 6h09
Objet : Re: [CES #16210] Re: Compromise between Approval Voting and Range Voting advocates
Brian Olson
May 16, 2017, 12:48:58 PM5/16/17
to electio...@googlegroups.com
I made a webapp for generating printable rankings ballots, shouldn't be hard to turn it around into ratings ballots
here's example output
make your own here:
On Tue, May 16, 2017 at 5:41 AM, 'Hélène Mayeur' via The Center for Election Science <electio...@googlegroups.com> wrote:
Looks like the ballot in the Vote de Valeur experiment in 2012. (http://www.votedevaleur.org/)Hélène Mayeur
To unsubscribe from this group and stop receiving emails from it, send an email to electionscience+unsubscribe@googlegroups.com.
--
You received this message because you are subscribed to the Google Groups "The Center for Election Science" group.
To unsubscribe from this group and stop receiving emails from it, send an email to electionscience+unsubscribe@googlegroups.com.
David Hollander
May 16, 2017, 5:25:06 PM5/16/17
to The Center for Election Science
If you would like to edit the mockup I previously posted, I have linked and attached the HTML code[1], with the necessary CSS rules embedded. Ovals are generated using the CSS 'border-radius' property. The "Vote de Valeur" website has an image which could possibly be used as a basis for generating the tabulation worksheet[2].
I believe it may be feasible in the future to create a "paper range voting election editor" using a single HTML file and javascript, where the ballot format data can be exported and imported using JSON, somewhat analagous to the website "keyboard layout editor"[3].
[1] https://pastebin.com/raw/g8CzHMiS
[2] http://www.votedevaleur.org/res/calculResultats.png
[3] http://www.keyboard-layout-editor.com/
I believe it may be feasible in the future to create a "paper range voting election editor" using a single HTML file and javascript, where the ballot format data can be exported and imported using JSON, somewhat analagous to the website "keyboard layout editor"[3].
[1] https://pastebin.com/raw/g8CzHMiS
[2] http://www.votedevaleur.org/res/calculResultats.png
[3] http://www.keyboard-layout-editor.com/
On Tuesday, May 16, 2017 at 11:48:58 AM UTC-5, Brian Olson wrote:
I made a webapp for generating printable rankings ballots, shouldn't be hard to turn it around into ratings ballotshere's example outputmake your own here:
On Tue, May 16, 2017 at 5:41 AM, 'Hélène Mayeur' via The Center for Election Science <electio...@googlegroups.com> wrote:
Looks like the ballot in the Vote de Valeur experiment in 2012. (http://www.votedevaleur.org/)Hélène Mayeur
Envoyé le : Mardi 16 mai 2017 6h09
Objet : Re: [CES #16210] Re: Compromise between Approval Voting and Range Voting advocates
I have attached and linked a mockup of a Range-5 ballot using the evaluative \ signed-integer scale in case it is of interest:
https://i.imgur.com/Xbpub8s.png
On Monday, May 15, 2017 at 9:04:23 PM UTC-5, parker friedland wrote:I put YouTube under a range of 3 (net approval voting) however right now YouTube's algorithm does not care about dislikes very much. In fact it slightly favors video's that have been disliked rather then not liked or disliked even when both are watched for the same amount of time:
(warning: skip to 0:10s because of this guy's really annoying intro)
https://www.youtube.com/watch? v=Om4c4ilT_6g
On Monday, May 15, 2017 at 6:41:47 PM UTC-7, parker friedland wrote:However when it comes to voting, I strongly prefer the evaluative scale {-2,-1,0,+1,+2} / {-1, 0, 1} over the Google scale. In fact I believe that in ranges of 5 and 3, this scale will decrease use of the Max/Min voter heuristics because 0 is a much more meaningful reference point then 1 and 3 are.
On Monday, May 15, 2017 at 6:31:38 PM UTC-7, parker friedland wrote:In fact huge corporations like google and Netflix seem to agree with me on that because their most common ranges are ranges of 5 (example: Google play store), 3 (Netflix, YouTube), and 2 (Facebook likes)
On Monday, May 15, 2017 at 6:26:15 PM UTC-7, parker friedland wrote:The more you expand the range, the most common ratings voters will assign to each candidate will become max, min, and 0 (assuming your using the evaluative scale). Heuristics do become a problem after that and after having at least 5 values to rate each candidate, the information gained as a result of adding more values becomes minuscule and it does still seem a little useful to provide voters with the values in between 0, max, and min. I believe net approval is the most reasonable because it cuts straight to the point by providing voters the scores that actually matter however a point can be made about token candidates that makes a 5 valued evaluative scale also pretty reasonable. I don't believe anything beyond that is reasonable.
On Monday, May 15, 2017 at 5:29:55 PM UTC-7, David Hollander wrote:On Monday, May 15, 2017 at 6:36:01 PM UTC-5, Warren D. Smith (CRV cofounder, http://RangeVoting.org) wrote:
> it does NOT particularly hurt score-voting style telephone polls, because you do not need to keep everybody else in mind as a pollee,
It is true that the pollee does need to keep in mind each candidate when stating a score. However, they do need to keep in mind the range of possible scores. The number of candidates on the ballot and the range of possible scores which may be assigned to each candidate are orthogonal considerations. With a large score range greater than Range-7, voters will be more likely to employ heuristics in order to determine the score they assign to each candidate. Heuristics such as the availability heuristic and anchoring-and-adjustment heuristic may introduce bias and decrease voter honesty. For an extremely large number range, political consultants will inform their candidate to verbally use the highest number on the scale as frequently as possible. This is in order to associate their candidate with this number in the minds of voters, so that voters use it as an anchor point when employing common numerical heuristics.
> With approval, your decision about where to place your personal "dividing line" for acceptable vs not, DEPENDS on who is in the race.
I agree. There is also a lower bound on the appropriate range. If the range is so small that it results in "lossy compression", then the voter is effectively dropping bits of information from their preference encoding. If they are forced to drop a large number of bits, they will want to consider how this affects the quality of the final results, in order minimize the perceptible loss of information.
> "digit span" test
Memorization and recall of a sequence may not be as relevant to score voting as identification of a unique value. There are many compression algorithms and strategies available for ordered sequences which would not be relevant for score voting. I believe one such strategy involves chunking subspans of digits into imaginary pictures which form a narrative. The number of chunking operations which voters should be required to perform to compress the possible range of numbers in range voting should perhaps be minimized, as chunking decisions may be performed using biased heuristics spread by political campaign consultants in order to distort results.
--
You received this message because you are subscribed to the Google Groups "The Center for Election Science" group.
To unsubscribe from this group and stop receiving emails from it, send an email to electionscien...@googlegroups.com.
--
You received this message because you are subscribed to the Google Groups "The Center for Election Science" group.
To unsubscribe from this group and stop receiving emails from it, send an email to electionscien...@googlegroups.com.
David Hollander
Jun 13, 2017, 5:36:43 PM6/13/17
to The Center for Election Science
I have uploaded a single-page FAQ for the range-5 + evaluative scale variant at the following domain name:
http://www.range5.org
I may prioritize the following improvements in the near future:
- add a "How are elections counted?" section with tabulation example
- add a "How are elections secured?" section with best practices for handling paper ballots
- add "What is vote splitting?" and "What is favorite betrayal?" sections
- add a more detailed references section with external links to articles on rangevoting.org or CES
- possibly provide an English translation of some of the content on the Vote de Valeur website
- add a contact email address
This website will most likely remain fairly minimalistic and redirect visitors searching for longer-form articles to other sites.
DH
http://www.range5.org
I may prioritize the following improvements in the near future:
- add a "How are elections counted?" section with tabulation example
- add a "How are elections secured?" section with best practices for handling paper ballots
- add "What is vote splitting?" and "What is favorite betrayal?" sections
- add a more detailed references section with external links to articles on rangevoting.org or CES
- possibly provide an English translation of some of the content on the Vote de Valeur website
- add a contact email address
This website will most likely remain fairly minimalistic and redirect visitors searching for longer-form articles to other sites.
DH
William Waugh
Jun 18, 2017, 7:36:47 PM6/18/17
to The Center for Election Science
On Sunday, May 14, 2017 at 12:41:41 AM UTC-4, parker friedland wrote https://groups.google.com/d/msg/electionscience/5FNFuX4mn6Q/mu9OEWjcFwAJ
With a voting strategy using random numbers, all ranges (having at least two distinct values) are equivalent in terms of the power available to a given voter to influence the outcome. All require the same amount of attention on how well the candidates are likely to do with the other voters.
With a voting strategy using random numbers, all ranges (having at least two distinct values) are equivalent in terms of the power available to a given voter to influence the outcome. All require the same amount of attention on how well the candidates are likely to do with the other voters.
The middle of the range is not neutral unless it is also the average of the scores you assign. For example if I vote Green 50, Blue 49, Dark Mare 0, and Trump -50, I am moving the total of Dark Mare downward with respect to the center of gravity of the other candidates, thus I am hurting her chances of being elected, thus 0 in this case represents weak opposition.
David Hollander
Jun 18, 2017, 10:16:48 PM6/18/17
to The Center for Election Science
If we were to conduct a marginal analysis of the impact of incremental adjustments to scores from an initial condition where all candidates start with a default score of zero, then a zero score could not be used to indicate opposition.
Dark Mare's chances of winning the election were only decreased because the voter decided to adjust the scores for Green and Blue higher in a show of support, not because they left her score at zero.
If the voter did not adjust scores and gave all candidates an equal score of zero, then zero could not possibly represent weak opposition, as the voter's impact on the election would be equivalent to not voting at all.
This is not dependent on how other voters decide to vote, but on how the voter decides to fill in their own ballot.
Additionally, single winner elections may not simply be zero sum contests, but also an empirical measurement of the popularity of policy combinations on which a candidate campaigned and a measurement of the political mandate they have for enacting these policies. It would make sense to use a zero-centered range voting even if there was only single candidate on the ballot, because this would act as a job performance review. If a politician was elected with a negative score, this would indicate that they have no real political mandate to lead, and incentivize others to run for office as challengers in the next election. I believe in the range-3 elections of the Republic of Venice which Warren describes on his site, a negative score would resort in a do-over and new round of elections, presumably because the electors viewed the ability of candidates to break past a pre-defined threshold as an important indicator of future job performance.
Dark Mare's chances of winning the election were only decreased because the voter decided to adjust the scores for Green and Blue higher in a show of support, not because they left her score at zero.
If the voter did not adjust scores and gave all candidates an equal score of zero, then zero could not possibly represent weak opposition, as the voter's impact on the election would be equivalent to not voting at all.
This is not dependent on how other voters decide to vote, but on how the voter decides to fill in their own ballot.
Additionally, single winner elections may not simply be zero sum contests, but also an empirical measurement of the popularity of policy combinations on which a candidate campaigned and a measurement of the political mandate they have for enacting these policies. It would make sense to use a zero-centered range voting even if there was only single candidate on the ballot, because this would act as a job performance review. If a politician was elected with a negative score, this would indicate that they have no real political mandate to lead, and incentivize others to run for office as challengers in the next election. I believe in the range-3 elections of the Republic of Venice which Warren describes on his site, a negative score would resort in a do-over and new round of elections, presumably because the electors viewed the ability of candidates to break past a pre-defined threshold as an important indicator of future job performance.
William Waugh
Jun 19, 2017, 10:26:51 PM6/19/17
to The Center for Election Science
I was wrong.
I believe that there is no way I can give some candidate the maximum score, another candidate the minimum score, and a third candidate such a score as to preserve her win/tie/lose status as would happen if I didn't vote, without my knowing the subtotals of how the other voters vote.
In https://groups.google.com/d/msg/electionscience/5FNFuX4mn6Q/LWjaiF_XCAAJ I insinuated that giving a candidate a score equal to the average of the other scores I give would result in her winning if she would have won without my vote, tying for the win if she would have tied for the win without my vote, and losing if she would have lost without my vote. I found a counterexample, so the insinuated statement is false.
I believe that there is no way I can give some candidate the maximum score, another candidate the minimum score, and a third candidate such a score as to preserve her win/tie/lose status as would happen if I didn't vote, without my knowing the subtotals of how the other voters vote.
We know that the maximum boost a voter can give a candidate is by assigning him the max score and all others the min score. We know that the maximum opposition a voter can give a candidate is by assigning him the min score and all the others the max. Not much can be said in between, about whether a vote helps a candidate, hurts him, or has neutral effect, except that if the voter gives all candidates the same score, it is neutral with respect to all candidates and that voter might as well have stayed home. So I think that there is no assured neutral stance toward a candidate (in terms of effect on the outcome) except the degenerate case.
David Hollander
Jun 20, 2017, 2:27:06 AM6/20/17
to The Center for Election Science
I believe a three candidate vote of (0, 0, 0) is only equivalent to a vote of (-2,-2,-2) or (+2,+2,+2) if we choose to make the assumption that elections are zero sum. If elections are not zero-sum, and the sum of scores is important, then any vote which is not (0,0,0) is still valuable to make. A (-2,-2,-2) vote will decrease the mandate of whoever is elected, and a (+2,+2,+2) will increase the mandate of whoever is elected. In a single candidate election, such as the confirmation vote on executive nominations held by the United States Senate, the sum determines whether a new candidate will be put forward who better addresses concerns. The recent confirmation of Attorney General Jeff Sessions might provide a good case study for the use of range voting in single-candidate elections. In multi-candidate elections without explicit rules requiring new candidates to put forward for negative sums, the perceived sum may still act as an important signal of the strength of public institutions and control the rate of investment in public vs private endeavors by actors within the voting area.
I would agree that any move from the initial position is non-neutral if the initial position was neutral and the new position is not equivalent. However in order for us to describe an election method as neutral, rather than a move as neutral, we might simply need to show that the election method provides a clear set of initial conditions, that these initial conditions are the same for each participant, and that the initial conditions do not favor a specific outcome.
If we were to introduce a bounded rationality model for voters which states that voters suffer from an 'availability' bias where the initial position taken will not be their optimal position, and an 'anchoring and adjustment' bias where the moves from their initial position will not be sufficient to reach their optimal position, then we might want the ballot instructions to describe a default position to take, and ensure that the average quantity of moves which voters must take to reach a satisfactory approximation of their optimal position from the default position is minimized.
I would agree that any move from the initial position is non-neutral if the initial position was neutral and the new position is not equivalent. However in order for us to describe an election method as neutral, rather than a move as neutral, we might simply need to show that the election method provides a clear set of initial conditions, that these initial conditions are the same for each participant, and that the initial conditions do not favor a specific outcome.
If we were to introduce a bounded rationality model for voters which states that voters suffer from an 'availability' bias where the initial position taken will not be their optimal position, and an 'anchoring and adjustment' bias where the moves from their initial position will not be sufficient to reach their optimal position, then we might want the ballot instructions to describe a default position to take, and ensure that the average quantity of moves which voters must take to reach a satisfactory approximation of their optimal position from the default position is minimized.
Reply all
Reply to author
Forward
0 new messages