Score Voting, but any candidate who gets a 0 (or isn't scored) by a majority of voters can't win (unless nobody can win)

(https://www.reddit.com/r/EndFPTP/comments/d7rssc/utility_lost_by_honest_voters_as_a_strategic/)

This allows majorities to more honestly differentiate between their preferred candidates and also support consensus candidates without the fear that they will end up with no utility at all. So in situations where a majority is split between two candidates, they can show their honest preference for either without letting the minority beat both on total points.

The two situations where this rule breaks down are when 1) voters cast protest votes for unviables, raising the bar for a majority and 2) a minority honestly more intensely prefers its candidates than the majority. The first could perhaps be solved by having a dynamic plurality requirement i.e. 40% instead of >50%, and may not even be a big deal to begin with for viability and quality, and the second is probably not a very likely or harmful situation overall, and may even make regular Score less viable than this.

This idea probably doesn’t have much room to grow with Score and STAR already out there, but perhaps it could be floated around for legislative decision-making or private elections, as a way of balancing utilitarianism with respect for the majority’s wishes.

One alteration is to instead eliminate candidates who don’t receive >50% support (more than a 5/10) from a majority; this could be called the “candidates require a majority of support from a majority” version. And a further alteration, to address situations where a minority more intensely prefers its candidates, is to only trigger the rule when a majority gives at least one candidate >50% support, but these complicate the method too much to be viable.

On first look, such procedures are a step away from the simplest simple score voting. They certainly make more work for the tabulators, thus make computer voting machines more attractive. I don’t put much faith in computer simulations of voting methods, but I do work through the possibilities in my head. I will think about this for about a week, and report back…

2 Likes

Since this is 100% failing all the criteria that the partisan proportional representation advocates care about, why not go to single member score and get back all the local representation?

This is a variation on single-winner Score. An equivalent in PR might be that a candidate can’t win if they are not positively scored by at least a majority of a Hare Quota Droop Quota? (possibly only looking at the ballots that still have significant amounts of weight), though I doubt it’d improve things much there.

I came up with a failure case: if there are two sides bullet voting their candidate that are <50% of the population, and there’s a candidate everyone agrees is 1/5, the 1/5 candidate wins. The fixes for this take Score further away from utilitarianism and viability unfortunately.

The simplest fix is to only disqualify candidates a majority explicitly scores 0; not scoring them is not enough. But in that case, it’s unlikely the rule would ever be triggered outside of small elections or decision-making situations.

It’d be interesting to do Score, but a majority can “approve” certain candidates who then can’t lose to any other candidates, even if other candidates have higher scores. That would probably fix the biggest flaws people pick with Score, since a majority can only lose if they don’t mind losing.

I think most of the people who oppose Score think that voters will tend to vote 10,0,0,0,…,0. If you want to “fix” Score, address that, despite the fact that it is largely a misconception.

A much more valid criticism would be that the chicken dilemma will be a major problem for majorities in Score, because they may split the vote to some extent; having an option to guarantee a majority-preferred winner (you score one of the candidates on your side a 10 and the other a 1, and approve/prefer both) would be a great fix to that issue.

In standard score, if a majority 0s a candidate, then that candidate can only win if no one has >=50% support anyway.

Anyway, here is an alternate solution to the problem you are proposing. It’s not great, but it might be editable into something useful.

  • Candidates with common interests who expect that they are likely to compete with each other for votes may agree to form an “alliance” (for example, a party, but not necessarily.)
  • Voters can score individual candidates, and or alliances. They may also give “internal scores” to a particular alliance, no greater than the external score they give the candidate.
  • The winning candidate shall be the one who meets the following criteria:
  1. There are no candidates outside their alliance with a greater external score.
  2. There are no candidates inside the alliance with a greater internal score who have not become ineligible to win by the previous criteria.

For example, there are candidates A,B,C,D. A,B,C form an alliance.
The ballots are listed in the form Candidate,External Score/Internal Score. The ballots are:
40 A 5/5 B 5/2 C 5/0 D 0/0 (Hardliner partisans, for alliance)
8 A 3/0 B 5/5 C 5/3 D 0/0 (Average partisans, for alliance)
3 A 2/0 B 5/4 C 5/5 D 0/0 (Moderate partisans, for alliance)
20 A 0/0 B 0/0 C 2/2 D 5/5 (Moderates, lean against alliance)
29 A 0/0 B 0/0 C 0/0 D 5/5 (“Partisans, against alliance”)
External standings:
C 295
B 255
D 245 (Can’t win, externally loses to B and C)
A 230 (Can’t win, externally loses to D)

Internal standings:
B 132 (Elected)
C 79

That said @rkjoyce is right to value simplicity of reform, since complications mean people won’t hear us out. So I won’t be advocating this system.

Not if the majority splits the vote due to chicken dilemma or simply indicating preference between their various candidates.

It really seems to me that if a majority were allowed to express either support or opposition to certain candidates, or perhaps following your alliance idea, could support or oppose certain alliances (i.e. a majority of Democrats indicate opposition to the Republican alliance, therefore that alliance can’t win even if some of their candidates have higher scores than everyone in the Democratic alliance), that’d be more than enough to limit the effect of a chicken dilemma; a voter would have to play dangerous games to simultaneously bullet vote their favorite but also try to oppose other candidates in their alliance in an attempt to make only their favorite a majority-supported candidate.

It could perhaps be automatically calculated whether a majority supports or opposes particular candidates i.e. if the majority of voters score three Democrats either a 5 or a 1, then even if two Republicans in the race get 40% support, the majority can be presumed to prefer any of the Democrats. So it’d be like STAR but able to capture preferences between all groupings of candidates, rather than only the two in the runoff. This probably would best be done as “Score but the highest-scored Smith Set candidate is guaranteed to win” perhaps with a condition that the Smith Set candidate receive at least a certain % of the Score winner’s support.