Navigation

    Voting Theory Forum

    • Register
    • Login
    • Search
    • Recent
    • Categories
    • Tags
    • Popular
    • Users
    • Groups

    Smith // Score

    Single-winner
    2
    10
    383
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply
    This topic has been deleted. Only users with topic management privileges can see it.
    • J
      Jack Waugh last edited by

      Has anyone reduced the explanation of how to tally Smith // Score to JavaScript or detailed pseudocode?

      Approval-ordered Llull (letter grades) [10], Score // Llull [9], Score, STAR, Approval, other rated Condorcet [8]; equal-ranked Condorcet [4]; strictly-ranked Condorcet [3]; everything else [0].

      L 1 Reply Last reply Reply Quote 0
      • L
        Lime @Jack Waugh last edited by Lime

        @jack-waugh said in Smith // Score:

        Has anyone reduced the explanation of how to tally Smith // Score to JavaScript or detailed pseudocode?

        Just find the Smith set (smallest set of candidates with no pairwise losses outside the set):
        https://electowiki.org/wiki/Maximal_elements_algorithms

        1 Reply Last reply Reply Quote 0
        • L
          Lime last edited by

          By the way, this can be done either with or without rescaling. The most common scalings would be to divide by either:

          1. The sup norm (i.e. normalize ballots), or
          2. The L2 norm/weight. This gives every voter an equal influence on the election, and also results in honest ratings if voters think all candidates in the Smith set will have similar scores (the zero-information setting).
          J 2 Replies Last reply Reply Quote 0
          • J
            Jack Waugh @Lime last edited by

            @lime, that's an interesting point. I think it makes good sense to rescale.

            Approval-ordered Llull (letter grades) [10], Score // Llull [9], Score, STAR, Approval, other rated Condorcet [8]; equal-ranked Condorcet [4]; strictly-ranked Condorcet [3]; everything else [0].

            1 Reply Last reply Reply Quote 0
            • J
              Jack Waugh @Lime last edited by

              @lime, but on the other hand, rescaling is not practical, because it would require reexamining the ballots. The people need precinct-summable systems to maintain security against fraud.

              Approval-ordered Llull (letter grades) [10], Score // Llull [9], Score, STAR, Approval, other rated Condorcet [8]; equal-ranked Condorcet [4]; strictly-ranked Condorcet [3]; everything else [0].

              L 1 Reply Last reply Reply Quote 0
              • L
                Lime @Jack Waugh last edited by

                @jack-waugh said in Smith // Score:

                @lime, but on the other hand, rescaling is not practical, because it would require reexamining the ballots. The people need precinct-summable systems to maintain security against fraud.

                Can be done in two passes. The first pass identifies the Smith set, and then a second pass identifies the Score winner.

                J 1 Reply Last reply Reply Quote 0
                • J
                  Jack Waugh @Lime last edited by

                  @lime, yeah, you could. There is a little bit more risk that the workers in one precinct get tired and keep everyone waiting.

                  Maybe a useful policy would say conduct the election with computers and the Internet, let the result go into effect, then verify everything by hand. It might be easier to check a proposed outcome than to compute it from scratch.

                  Approval-ordered Llull (letter grades) [10], Score // Llull [9], Score, STAR, Approval, other rated Condorcet [8]; equal-ranked Condorcet [4]; strictly-ranked Condorcet [3]; everything else [0].

                  1 Reply Last reply Reply Quote 0
                  • L
                    Lime last edited by Lime

                    @jack-waugh said in Smith // Score:

                    @lime, yeah, you could. There is a little bit more risk that the workers in one precinct get tired and keep everyone waiting.

                    Maybe a useful policy would say conduct the election with computers and the Internet, let the result go into effect, then verify everything by hand. It might be easier to check a proposed outcome than to compute it from scratch.

                    The Essential set almost-always has 3 candidates, so the method can be made 3rd-order summable except when the election gets ridiculously close.

                    1 Reply Last reply Reply Quote 0
                    • J
                      Jack Waugh last edited by

                      I need to study the Smith set more and the algorithms, or at least one of them.

                      Approval-ordered Llull (letter grades) [10], Score // Llull [9], Score, STAR, Approval, other rated Condorcet [8]; equal-ranked Condorcet [4]; strictly-ranked Condorcet [3]; everything else [0].

                      1 Reply Last reply Reply Quote 0
                      • J
                        Jack Waugh last edited by

                        We who want to eliminate vote-splitting and spoiler effects have grounds to choose a system that the public can easily understand. Does anyone think this system passes muster in that regard?

                        Approval-ordered Llull (letter grades) [10], Score // Llull [9], Score, STAR, Approval, other rated Condorcet [8]; equal-ranked Condorcet [4]; strictly-ranked Condorcet [3]; everything else [0].

                        1 Reply Last reply Reply Quote 0
                        • First post
                          Last post