Approval Score Voting (A-SV)

I propose this voting system. Tell me if it already exists under another name.

Use votes with range [0,9] (or [0,5]).

Is there at least one candidate who has obtained a rating greater than or equal to 9 in more than 50% of the votes?

  • Yes: the candidate with the highest sum of points, among those satisfying the condition, wins.
  • No: I repeat the procedure, decreasing by 1 the value of the rating to be checked.
    The new condition will be: is there at least one candidate who has obtained a rating greater than or equal to 8 in more than 50% of the votes?..

This process sooner or later ends by returning a winner.

Variants (extra)
This voting logic can also be extended to other methods, such as rankings:
A 3-position ranking is used, recursively.
1st vote ranking: A [1st] B [1st] C [2nd] D [3rd] E [3rd]
2nd vote ranking: A [1st] B [2nd] C [1st] D [1st] E [2nd]
The 1st ranking of votes is used, and depending on the procedure, possibly the 2nd ranking is also used (in case the 1st is not enough to find a clear winner).
This succession of rankings can be used to introduce the concept of cardinality in ranking votes as well.

This variant can be useful on the web regarding tier lists in which in a certain position there are many candidates who are not really equal, so you could create an additional (2nd) tier list for each position of the 1st tier list.

More than a voting system, it is a form in which the vote can be written.

See majority judgement, graduated majority judgement, and ER-Bucklin.

1 Like

Seen, they are similar but use rankings, which represent the order but not the distance of the candidates.

Ex. vote with range [0,9]:
A [9] B [2] C [0]
in the first step the vote is treated as:
A [X] B [-] C [-]
but this is also true in the 2nd, 3rd, 4th… steps because between A and B there is a distance of 7.
Rankings don’t work like that, but thanks for pointing me out those similar methods.

Majority Judgment uses ratings, not rankings. It is a form of median rating.


It seems like it is susceptible to vote splitting.

Say there are 5 candidates on the right, and 1 on the left.

Voters on the right will lower ratings for some of the candidates, so that they can express a difference between them. Voters on the left will all give their single candidate a 9. The candidate on the left will gain a significant advantage.

The result of this is that parties will continue to hold nominations (e.g. based on primaries) to eliminate candidates prior to the general election.

I do encourage you to put it into a CodePen and we can hammer on it with sample ballot sets. Seems like it would be easy to code up and put it into one of our existing CodePens.

When in definition spoke of grades, I thought he meant values that at least once had to be used (as rankings), but I misunderstood.

If the left alone doesn’t have more than 50%, then the voting method becomes like SV (which effectively fails the IWA, which seems like the problem you present).
In this method I was trying to satisfy the monotony, so it is normal that the IWA is not satisfied.

Particularly @robla 's hypothetical
49 A[100] B[52] C[0] D[0]
1 A[50] B[51] C[100] D[0]
49 A[49] B[0] C[0] D[100]