3 ms·

Wikipedia has this to say (https://en.wikipedia.org/wiki/Ranked_voting https://en.wikipedia.org/wiki/Ranked_voting): Dowdall's method assigns 1, 1⁄2, 1⁄3... po

Wikipedia has this to say (https://en.wikipedia.org/wiki/Ranked_voting https://en.wikipedia.org/wiki/Ranked_voting): Dowdall's method assigns 1, 1⁄2, 1⁄3... points to the 1st, 2nd, 3rd... candidates on each ballot, then elects the candidate with the most points. Ranked voting systems vary dramatically in how preferences are tabulated and counted, which gives each one very different properties. Ranked voting systems are usually contrasted with rated voting methods, which allow voters to indicate how strongly they support different candidates (e.g. on a scale from 0-10).[1] Rated voting systems use more information than ordinal ballots; as a result, they are not subject to many of the problems with ranked voting (including results like Arrow's theorem). That's a bucket full of problems. I remember that I had a similar problems years ago when I wanted to consolidate several lists about the "daily usefulness" of Chinese characters. How much does it weigh when you put A, B, C in that order? is A=1/1, B=1/2, C=1/3? or maybe A=0, B=-1, C=-2? To quote the above, "Ranked voting systems vary dramatically in how preferences are tabulated and counted". This should be a red flag! I'm happy to see that alternatives to first-past-post are tried out in some locations such as Alaska, Maine and Australia, but why ranked voting? Ranked voting is only good when you think of yourself as a sole, lonely voter. You put the candidates into this order: A, B, C. But there will be other voters with other preferences. Let's say you're three people and the rankings are ABCD, EBCD and FBCD. Now you have three favorites A, E, F that are preferred choices, and there's that also-ran B who only was second best. But B got three second-best votes whereas each of A, E, F only got one vote. So answer my question: how many second, third, fourth votes will it take to surpass a given number of first votes? If there's another ballot in the same election with GCBD on it, now you have B with 3x2nd and 1x3rd place. You cannot add, subtract, multiply or divide these numbers. There's no clear answer (although practical procedures can and have been proposed and used) and that bugs me. The entire thing looks like a game of Bridge, Skat or Poker to me, full of justifiable but ultimately very arbitrary rules, hard to explain and tricky to get right. https://en.wikipedia.org/wiki/Condorcet_method https://en.wikipedia.org/wiki/Condorcet_method says that much to me. One also has to account for unranked candidates and decide whether it should be possible to give one rank to several candidates, and then whether (as done in sports) the next rank must be left empty. It just goes on and on, the list of options to do the ballot and to do the counting is truly mind boggling. This is the reason I've come to much prefer preference voting over ranked voting. Ranked voting is much too complicated for general use in a general election whereas preference voting is simple and clear and the number of "design choices" as it were is drastically reduced when compared to ranked voting.

JoshTriplett 2y ago
"preference voting" and "ranked choice" are largely used as synonyms, as both involve an ordered list of preferences. When I mentioned writing out a list of numbers in order, I was talking about ballot UX design here. The ordinary case of preference voting is extremely explainable: if more people prefer A to B than prefer B to A, A beats B; pick the candidate that beats everyone else. "It's like running a bunch of head-to-head elections all at once." They're also really easy to tally into a table: "A ranked above B" is a point for A over B; "B ranked above A" is a point for B over A; A and B ranked equal relative to each other doesn't add anything to either of those two cells. The corner cases like cycles almost never come up. And even then, it's relatively easy to explain a high-level understanding of the most common methods: "if there's a loop of people who each beat the next one in the loop, ignore the one with the smallest margin of victory". If A beats B in a landslide, B beats C in a landslide, and C very narrowly beats A, A wins. Also: > Let's say you're three people and the rankings are ABCD, EBCD and FBCD. Every ballot implicitly ranks every option, and the usual assumption is that if someone doesn't even care enough to list a candidate, they prefer every candidate they did list over any candidate they didn't list. (There are UX design ideas that could let someone say "this candidate is last" without ordering every candidate, but that adds complexity.) So these ballots are: A > B > C > D > E = F E > B > C > D > A = F F > B > C > D > A = E Writing it out that way shows that B beats A by 2:1, B beats E by 2:1, B beats F by 2:1, B beats C by 3:0, and B beats D by 3:0. No corner cases or cycles here; a majority of people prefer B to every other candidate.

JoshTriplett 2y ago

"preference voting" and "ranked choice" are largely used as synonyms, as both involve an ordered list of preferences. When I mentioned writing out a list of numbers in order, I was talking about ballot UX design here. The ordinary case of preference voting is extremely explainable: if more people prefer A to B than prefer B to A, A beats B; pick the candidate that beats everyone else. "It's like running a bunch of head-to-head elections all at once." They're also really easy to tally into a table: "A ranked above B" is a point for A over B; "B ranked above A" is a point for B over A; A and B ranked equal relative to each other doesn't add anything to either of those two cells. The corner cases like cycles almost never come up. And even then, it's relatively easy to explain a high-level understanding of the most common methods: "if there's a loop of people who each beat the next one in the loop, ignore the one with the smallest margin of victory". If A beats B in a landslide, B beats C in a landslide, and C very narrowly beats A, A wins. Also: > Let's say you're three people and the rankings are ABCD, EBCD and FBCD. Every ballot implicitly ranks every option, and the usual assumption is that if someone doesn't even care enough to list a candidate, they prefer every candidate they did list over any candidate they didn't list. (There are UX design ideas that could let someone say "this candidate is last" without ordering every candidate, but that adds complexity.) So these ballots are: A > B > C > D > E = F E > B > C > D > A = F F > B > C > D > A = E Writing it out that way shows that B beats A by 2:1, B beats E by 2:1, B beats F by 2:1, B beats C by 3:0, and B beats D by 3:0. No corner cases or cycles here; a majority of people prefer B to every other candidate.