Possible misrepresentation of STAR on Wikipedia


#23

Yes, strategic voting can be a concern in score voting… but not a 2PD level concern. When all voters are strategic in score voting, it degrades to approval voting, and when all voters are strategic under approval voting (under reasonable circumstances) a Condorcet winner will always win.

https://rangevoting.org/AppCW.html

https://rangevoting.org/CondAppConflict.html

Obviously, this is assuming that everyone is maximally strategic computers. The real question is, what happens when voters just think they are being strategic. If a significant chunk of the electorate think that bullet voting is the best strategy, then they would vote under honest plurality voting, which is very differently then strategic voting and such voters may actually get rid of the 2PD even faster then normal approval or score voting because candidate cloning makes honest plurality elections very unpredictable and increases the likelihood that less mainstream parties will win. Thus while strategic voting may be a concern in approval and score, it is not a 2PD concern.


#24

“voters are strategic under approval voting (under reasonable circumstances) a Condorcet winner will always win.” -ParkerFreidland

The thing is that the strategic incentive advantages some candidates over others. There’s no strategic advantage to giving a higher score to an underdog you think is pretty good. The incentive is to score your lesser evil higher. The Condorcet winner of actual ballots is not the same as the VSE or best winner from honest ballots. The honesty issue could lead to an accuracy issue where the system plays favorites.

“Why not just stop worrying about rectitudinous “honest” voters?” -rkjoyce “At some point it will become obvious that people must vote strategically if they are ever to defeat the great pirates”

I’m not blaming the strategic voters, of course people will and should vote strategically if that’s what they have to do to get a fair shake. The point is that we could have a system where that isn’t necessary because an honest vote is just as strong!


#25

Do you mind making a ballot vs. reality example for Score and Approval to show how they devolve into 2PD?


#26

There’s no strategic advantage to giving them a lower score either.

And to score your lesser good lower.


#27

We define an honest vote as one that accurately respects your true preferences. (Approval and Score/STAR are slightly harder to define, but if you prefer X over Y, scoring X=2 and Y=5 is clearly dishonest.)
You could argue that the word is wrong, but I see no reason right now to use a different word.

That was basically what I was arguing. I still think that Score Voting should be thought of as casting one “vote” giving 0 to 10 points to each candidate. But that is just terminology; the systems are the same.

There is no second election that voters have to attend. Instead, the Runoff is done automatically, with your full vote going to the candidate you scored higher. (If you rated them the same, you do not get a vote in the final round – but you have little to complain about since you deliberately decided not to differentiate two candidates you probably do not prefer one over the other anyway,)

This criticism, however, is legitimate. I am not sure which one is worse – simple Score voters may tend to Approval votes, but STAR could elect a polarizing pirate.


#28

I feel like CES and EVC are like candidates in the same political party competing in a primary election (whereas IRV is the opposite party). We both acknowledge each other as pretty good but still would rather come out on top.

Worst cases:

  • Score can lead to 2PD if people keep voting F1 and F2 at opposite ends of the scale, and no other candidate gets enough Score to bypass the front runners.
  • STAR can lead to 2PD if:
    1. the main parties run two at a time, to increase their chances of both (a) making the Runoff and (b) having both make the Runoff (= autowin for that party), AND also
    2. third parties are slow to come along, so that one of the polarizing front runner seats always has an absolute majority and autowins

Neither one is as bad as IRV.

Only those where (1) they genuinely think a front runner is the best candidate AND (2) there are no third parties that they consider reasonably decent enough to score above the main rival.


#29

AssetVotingAdvocacy: “Do you mind making a ballot vs. reality example for Score and Approval to show how they devolve into 2PD?”

Example: Bush vs Perot vs Clinton. 1994. Oversimplified.
Perot voter: Honest Ballot- Bush 2, Perot 5, Clinton 0.
Strategic Ballot- if you don’t think Perot can win: Bush 5, Perot 5, Clinton 0.

Bush voter Honest Ballot- Bush 5, Perot 2, Clinton 0.
Strategic Ballot- if you don’t think Perot can win: Bush 5, Perot 2, Clinton 0.
Strategic Ballot- if you do think Perot can win: Bush 5, Perot 0, Clinton 0.

Clinton Voter Honest Ballot- Bush 0, Perot 2, Clinton 5.
Strategic Ballot- if you don’t think Perot can win: Bush 0, Perot 2, Clinton 5.

The strategic voting benefits the candidate who is deemed viable. But, what if it turns out Perot was actually viable based on honest votes, but he was not deemed viable by the media? He does not get that same advantage. Most Perot voters would score Perot honestly with a 5, but those who like him somewhat but don’t think he could beat Bush don’t give a higher score than he deserves.

Our current system gives a huge advantage to those that the media deems viable because of strategic voting to avoid the spoiler effect. Score/Approval solves the spoiler effect, but to a much lesser extent, strategic voting still advantages candidates who the media deem viable, ie the establishment and those that raise the most money.

Is this phenomena enough to create or entrench 2PD? For score, maybe not, but it doesn’t help. For Approval, it’s the main concern for me…


#30

I’m going to be honest here: if Bush and Clinton voters don’t think Perot “deserves” higher than a 2, then he shouldn’t win under any system. Period. If he deserves higher, odds are he wins. Your concern seems to be more on majority rule than the best winner, and that concern may not be best answered by cardinal systems at all.


#31

Let’s break down that part about STAR.
Point #1. That scenario depends on the most popular party’s candidates being clones who are almost exactly tied. In that case one of those candidates deserves to win. (Same as the winner under Score Voting)

But if both individuals actually do want to win, then they have a huge incentive to try and differentiate themselves. If the candidates aren’t clones, then the one who better speaks to the people wins. In this scenario, which two parties are in control may not change, but the voters would be able to actually shift the parties closer to their base. End result. At least one of the major parties does represent the will of the people and the candidate who deserves to win, wins.

Point #2. If there aren’t good 3rd parties that people like then it’s not the voting systems fault that there would be only 2 parties. In that scenario, the candidate who deserves to win, wins.

But, STAR actually does a lot to combat polarization! Candidates who can represent voters beyond their base can earn points from additional voters that add to their total. That puts polarizing candidates at a big disadvantage.


#32

The strategic voting can benefit the candidate who is deemed viable. But it can also hurt the candidate who is deemed viable.

Suppose that a voter would honestly rate candidate A 5 stars, candidate B 4 stars, candidate C 0 stars. If the front runners where A and C, the voter would approve of A and B, since the voter certainly wouldn’t regret having approved of B if the winner ended up being C, however if the front runners were A and B, then the voter should only approve of A so that a B victory doesn’t prevent A from winning.

Strategic voting incentivizes voters to give a max score to one of the front runners, but it also incentivizes voters to give a min score to the other front runner.


#33

If both candidates are “viable” then that strategy doesn’t illustrate a net win or lose for “viable” candidates, as opposed to non “viable” candidates. It averages out, helping one “viable” candidate while hurting the other “viable” candidate.

What I’m trying to look at is if the system favors either those deemed viable or those that aren’t. Ideally it shouldn’t play favorites either way.