Navigation

    Voting Theory Forum

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

    SACRW2: Score And Choose Random Winner from 2 complementing methods

    New Voting Methods and Variations
    6
    16
    975
    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.
    • multi_system_fan
      multi_system_fan last edited by multi_system_fan

      I would be interested in a single winner winner systems that has these basic requirements:

      • start with a score ballot
      • has two independ and different methods to suggest one or more winners
      • if one method produces less suggested winners then the other then that method gets an extra round to add suggested winners until both methods have the same number of winners
      • ends with random picking a ultimate winner from the combined set of suggested winners (lets make that symbolical and festive: a little schoolchild chooses blind)
      • the winner is announced

      what two methods would you propose and why?
      as an example I'll give one:

      1 https://electowiki.org/wiki/Smith_set and
      2) https://rangevoting.org/ with the set of winners that have the maximum SUM of all scores.
      This would combine some aspects of ranking and rating without predictable results for strategizing. So (succesfull) strategizing would not happen much. Also importantly the randomization would be perceived by normal citizens (as non voting-experts) to be a fairly honest choice between two different kind of winners which both have a fair case to say they are legitimate and have a mandate. It could also lead to a bit more humbleness in the winner instead of arrogance because the winner would know he had an element of just luck on their side.

      ranked-irv[1] ranked-condorcet[2] ranked-borda[0] cardinal-condorcet[10] cardinal-median[4] cardinal-alt[10] star[6] approval[5] score[4] choose-one[1] minimax-TD[10]

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

        I favor your idea in general.

        In addition to the procedures you mention, Majority Judgment and STAR could be added. Why not use four methods?

        I particularly like your idea for the ceremony featuring a child making the random selection. I think it would appeal to citizens' emotions and make them feel solidarity around the overall method.

        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
        • multi_system_fan
          multi_system_fan last edited by multi_system_fan

          after a few adjustments I would change this to:

          RWLCP: Random Winner from a List of Complementing Philosophies

          A hybrid design for a single winner winner systems that combines some aspects of different voting-philosophies without predictable results for strategizing. So (succesfull) strategizing would not happen much. Also importantly the randomization would be perceived by normal citizens (as non voting-experts) to be a fairly honest choice between different kind of winners which all have a fair case to say they are legitimate and have a mandate. It could also lead to a bit more humbleness in the winner instead of arrogance because the winner would know he had an element of just luck on their side.

          steps:

          • start with a ballot that's suited to gather all needed info
          • use independ and different methods to suggest a single winner:
            1. MARS voting (Mixed Absolute Relative Score)
            2. Reciprocal Score voting
            3. SODA voting (Simple Optionally Delegated Approval)
            4. Rangevoting
            5. STAR voting
          • ends by picking the ultimate winner from the combined set of suggested winners (lets make that symbolical and festive: a little schoolchild chooses blind)
          • the ultimate winner is announced

          ranked-irv[1] ranked-condorcet[2] ranked-borda[0] cardinal-condorcet[10] cardinal-median[4] cardinal-alt[10] star[6] approval[5] score[4] choose-one[1] minimax-TD[10]

          1 Reply Last reply Reply Quote 0
          • Marylander
            Marylander last edited by

            In order for such a system to effectively deter strategic voting, the best strategies in the component systems used need to be sufficiently different. Otherwise common vulnerabilities still can be exploited. The most useful strategy in Score and Condorcet is usually exaggeration, so I am skeptical that the two systems effectively complement each other.

            There is also a risk that mixing many systems in this way could compound their strategic vulnerabilities rather than offset them.

            Strategic voting in Condorcet methods often attempts to increase the size of the Smith set and force cycles. The same strategies could be used with this system since the size of the Smith set also indicates how many winners the Score-based methods are expected to return. Random selection can be exploited by running many similar candidates, since when the Smith set contains multiple candidates, cloning a candidate in the Smith set would cause the clone to be added to the Smith set.

            Score Sorted Margins[100]; STAR[90]; Score[81]; Approval[59]; IRV[18]; FPTP[0]

            1 Reply Last reply Reply Quote 1
            • Sass
              Sass last edited by

              You'll never get lay people on board with a blatantly non-determinative method. If a blatantly non-determinative method was implemented, people wouldn't vote because they would feel, even more so than they already do with FPTP, that their vote doesn't matter.

              rob J Psephomancy 3 Replies Last reply Reply Quote 0
              • rob
                rob Banned @Sass last edited by

                @Sass Agree that anything with randomness is bad. Can't speak to people's incentives to vote though. Having some degree of randomness doesn't necessarily make their vote "matter" less, at least not from a game theoretical point of view. Still I think anything with true randomness would never fly in any real world political election.

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

                  @Sass said in SACRW2: Score And Choose Random Winner from 2 complementing methods:

                  You'll never get lay people on board with a blatantly non-determinative method.

                  On what grounds you come to that prediction? There is randomness in filling jury pools, and what juries do is extremely important as innocent people could be punished by the mechanism of the state. Moreover, people who I suppose have a libertarian bent, or at least, a mistrust of officialdom, keep suggesting, on Facebook, sortition to fill legislative seats and executive office in gov't.

                  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 1
                  • Psephomancy
                    Psephomancy @Sass last edited by Psephomancy

                    @Sass said in SACRW2: Score And Choose Random Winner from 2 complementing methods:

                    You'll never get lay people on board with a blatantly non-determinative method.

                    Why not? If the two methods elect the same person almost all of the time, while choosing between them randomly makes strategy impossible, people will understand.

                    I've often wondered about such a system, and if it provides a workaround for Gibbard's theorem. Design two voting systems that behave consistently if people are honest, but behave oppositely if people are strategic, and choose between them randomly, so that no one can predict whether a given strategy would help them or hurt them. I don't think you can pick two existing methods, though, I think you need to specifically design them to be contradictory.

                    Sass 1 Reply Last reply Reply Quote 0
                    • Sass
                      Sass @Psephomancy last edited by

                      @Psephomancy I can't speak on behalf of other countries, but Americans are so used to voting strategically that they're having trouble understanding the concept of a voting method that allows honest voting, let alone one that encourages it. The Equal Vote Coalition was a bit bewildered when folks we're only scoring two or three candidates on STAR ballots in some early tests. The voters just had trouble internalizing the idea that they could support multiple candidates. If you try to tell voters you have a method that picks a random winner if they vote strategically, they won't understand it and the media will latch onto "random winner" and ignore everything else. I support systems that incentivize honest voting, but that can be done with many existing methods that don't rely on randomness.

                      1 Reply Last reply Reply Quote 0
                      • Sass
                        Sass @multi_system_fan last edited by

                        @multi_system_fan https://electowiki.org/wiki/Smith//Score
                        Good luck explaining the Smith Set to lay voters.

                        rob 1 Reply Last reply Reply Quote 1
                        • rob
                          rob Banned @Sass last edited by

                          Introducing randomness into an election is a bad idea on so many levels. For instance, what if you have a recount? But just the general idea of it will be, and should be, offensive to many. This is not the same as picking a jury, obviously. (a jury is supposed to be random)

                          I'm sure voters can wrap their heads around the idea of a system that doesn't punish honest/sincere voting. Some will still try to vote strategically for a while, but those voters will essentially average out and not cause a lot of harm. The main point is that a system that favors honest voting.

                          I suspect the reason people only scored a subset of the candidates (in STAR elections, I guess) is that they didn't know anything about the ones they didn't give a score to. That's exactly what I'd do... just give them all an implicit zero. Even though I fully understand the idea of supporting multiple candidates.

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

                            @rob said in SACRW2: Score And Choose Random Winner from 2 complementing methods:

                            honest/sincere voting

                            Please consider the following system. The voters submit 32 hex digits as their vote. The tally sorts the votes, concatenates them together, and computes the SHA 256 hash of the whole. It interprets this result as an integer and divides the count of candidates into that. The remainder is used to indicate the winner.

                            Which votes are sincere?

                            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].

                            rob Marylander 2 Replies Last reply Reply Quote 0
                            • J
                              Jack Waugh last edited by

                              I misread your original post. I was thinking, collect rating ballots, use methods each of which is designed to pick a single winner, and choose at random from among the winners that the methods identify.

                              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
                              • rob
                                rob Banned @Jack Waugh last edited by

                                @Jack-Waugh It sounds like you simply described a random voting system with the 32 bit hex thing. Not sure what you're after here. I don't think your vote can be sincere when the voter doesn't know how to do anything other than put a random number down.

                                Regardless, I'm saying no one is going to accept a voting method with randomness in it as you describe, even with your more reasonable options. It is a non-starter.

                                That said, I think several "reasonable" methods are hard to say what a sincere vote is, Approval being the most obvious one. But a ranked method it is pretty straightforward to determine what a sincere vote is.

                                1 Reply Last reply Reply Quote 1
                                • Marylander
                                  Marylander @Jack Waugh last edited by

                                  @Jack-Waugh I think it is difficult to define a sincere vote as something completely isolated from its effects on the outcome, since I think that those effects are what give votes "meaning". Imagine a Score voting system where whoever had the fewest points wins. A "sincere" vote under this system might be the exact opposite of a sincere vote under standard Score voting.

                                  Under a system where the effects of your vote are be completely unpredictable, like the hashing one you described, I don't think your vote is capable of expressing anything about your preferences, since there is no way to tell in advance what any vote actually does.

                                  Score Sorted Margins[100]; STAR[90]; Score[81]; Approval[59]; IRV[18]; FPTP[0]

                                  rob 1 Reply Last reply Reply Quote 0
                                  • rob
                                    rob Banned @Marylander last edited by

                                    @Marylander said in SACRW2: Score And Choose Random Winner from 2 complementing methods:

                                    I think it is difficult to define a sincere vote as something completely isolated from its effects on the outcome, since I think that those effects are what give votes "meaning"

                                    True, but with ranked methods, I think we can pretty easily define "sincere" to mean that you ranked them according to your actual preferences. Likewise under plurality, if I preferred Ralph Nader to Al Gore (in 2000), but voted for Gore because I thought Nader didn't have a chance, and preferred Gore to Bush... that's an insincere, strategic vote.

                                    One shortcut to deciding what is a sincere vote is if you actually would vote the same way if you truly had no idea which of the candidates is more likely to win. If so, that is (in all cases I can think of) a sincere vote.

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