Timeline for 2013 Community Moderator Election
Current License: CC BY-SA 3.0
24 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 3, 2013 at 17:26 | comment | added | Joseph O'Rourke | I used to manage our faculty elections, and I wrote my own STV software to support the elections. I find it an eminently fair system. I mean no disrespect to Henry, who obviously has thought about the issue carefully. But I don't see this as a significant issue. Let us see how it goes and consider alternatives if we find it does not sit well with us after the election. | |
| Oct 2, 2013 at 22:08 | comment | added | Vidit Nanda | Wow, I guess holding elections in a crowd with intimate knowledge of voting paradoxes is a bit like deciding what to eat for dessert in a diabetic community. | |
| Oct 2, 2013 at 17:38 | comment | added | Kaveh | (On philosophical grounds regarding Arrow's theorem, I think IIA is not realistic as studies in behavioral psychology show that even for one person the existence of even suboptimal alternative have an effect on the resulting decision.) | |
| Oct 2, 2013 at 17:27 | comment | added | Kaveh | I don't think we want arguments for their own sake but for some goal. I kind of feel that the goal is to maximize the total satisfaction of the participants about the outcome where all users are considered to be equals and I feel that STV does a good job in this respect. | |
| Oct 2, 2013 at 12:40 | comment | added | Henry Cohn | Using OpenSTV and publishing anonymous votes is a responsible way of handling STV voting, but it addresses a side issue (will STV be done correctly, rather than will it give a sensible result). The usual justification for STV is that voters who are used to plurality voting will not trust any method that looks different, so better methods are not even on the table and the only question is STV vs. plurality. That may be true for citizens voting in elections, but it shouldn't be true for mathematicians. | |
| Oct 2, 2013 at 12:31 | comment | added | Henry Cohn | In other words, the point is not just to select a winner, but to justify a winner to the community, and this is where STV falls apart completely. In simple cases it does fine (like every sane voting system), but in complicated elections, nobody can understand or predict what's going on, and that makes the results meaningless. These contentious cases are exactly where you want a strong argument for what, if anything, these results have to do with the will of the community. | |
| Oct 2, 2013 at 12:21 | comment | added | Henry Cohn | "Predictability" is probably not the right word, but I mean something absolutely essential. The only purpose of a voting system is to make an argument for how the voters' opinions should be integrated to arrive at a result. No way of doing this is perfect, but we still want a method that is intellectually compelling. Approval voting, range voting, and Borda count all have clear, compelling motivations. E.g., "these candidates won because they got the most approvals" or "these candidates won because they fared the best in pairwise comparisons". | |
| Oct 2, 2013 at 7:53 | comment | added | Michael Greinecker | In a sense, the intransparency of STV is its only positive quality: It is very hard to figure out how to vote strategically. | |
| Oct 2, 2013 at 5:33 | history | edited | Kaveh | CC BY-SA 3.0 |
added 86 characters in body
|
| Oct 2, 2013 at 5:31 | comment | added | Kaveh | SE uses an open-source software and publishes the anonymized votes after the election so people can recheck them. I don't think predictability is needed or even a desired property. As you mentioned all voting systems have some issues and critics. | |
| Oct 2, 2013 at 5:25 | history | edited | Kaveh | CC BY-SA 3.0 |
added 71 characters in body
|
| Oct 2, 2013 at 2:29 | comment | added | Joel David Hamkins | I would suggest that you make a post with such kind of arguments to meta.SO with a proposal to change the voting mechanism for all future SE elections. You could also post one answer per voting system, and see how the support goes... | |
| Oct 2, 2013 at 2:26 | comment | added | Joel David Hamkins | Thanks very much for your informative elaboration of your remarks! | |
| Oct 2, 2013 at 2:25 | comment | added | Henry Cohn | Sorry for the incredibly lengthy comments, but I figured I might as well give a more detailed outline of the background. (Hopefully those who don't want to read this won't expand the comments.) | |
| Oct 2, 2013 at 2:24 | comment | added | Henry Cohn | On the other hand, it's by far the most complicated method (in its iterative dynamics), and these sorts of complications seem problematic in a voting system. Furthermore, preserving plurality voting's primary role for first-place votes is not a good thing. If you're going to ask voters for more information than that, you should try to take it into account in as principled a way as you can, rather than gradually letting some of it leak into your results as a patch for plurality voting. | |
| Oct 2, 2013 at 2:22 | comment | added | Henry Cohn | STV is basically designed to be as close to plurality as possible. I.e., you start out by counting only first-place votes, but you iteratively adjust things as you eliminate candidates. In certain types of elections, this fixes problems with spoiler candidates (who will quickly get eliminated and play no further role). Furthermore, it can be politically more tenable since it feels like a modification of plurality voting, rather than an entirely different method. | |
| Oct 2, 2013 at 2:19 | comment | added | Henry Cohn | The system everyone hates is plurality, where only first-place votes count at all (as in U.S. presidential elections). Just about anything is better than that, since plurality has terrible issues with wasted votes, similar candidates splitting their support and losing to less popular candidates, etc. | |
| Oct 2, 2013 at 2:17 | comment | added | Henry Cohn | One possibility is Borda count: with n candidates, you give n-1 votes to your top candidate, n-2 to the next, ..., down to 0 to the last. Then these votes are totaled. Of course one can adjust the weightings, but this is a particularly nice choice (it sums up the number of pairwise comparisons each candidate wins). See the book Chaotic Elections by Saari for arguments in favor of the Borda count. | |
| Oct 2, 2013 at 2:15 | comment | added | Henry Cohn | Arrow's impossibility theorem says that there's no ideal method to combine voter rankings into a societal ranking. (Ideal meaning three properties. Unanimity, meaning if everyone prefers X to Y then so does society. Independence of irrelevant alternatives, meaning adding Z as a candidate cannot change the relative rankings of X and Y. And no dictatorship, meaning there isn't just one person who single-handedly determines the societal rankings.) This is bad news for democracy, and it means when there are more than three alternatives we have to make some compromises in the voting method. | |
| Oct 2, 2013 at 2:12 | comment | added | Henry Cohn | One philosophical objection to approval voting is how you compare what approval means between two different people. There's no absolute standard. For a different approach, you could ask everyone to rank the candidates. The idea that is preferring X to Y has a clear meaning independent of the voter, in a way that approval voting does not. The drawback is that it makes voting more cumbersome, and it's not clear how to tally the votes. | |
| Oct 2, 2013 at 2:10 | comment | added | Henry Cohn | I'd probably go with approval voting: vote for the subset of candidates you think would make good moderators, and the candidates who get the most votes win. That's by far the simplest system in just about every respect (implementation, explanation for voters, understanding the results). One drawback is that it doesn't take into account intensity of feelings; you can use extensions like range voting for this, but I'm not convinced it's important here. | |
| Oct 2, 2013 at 2:00 | comment | added | Joel David Hamkins | Would it be possible for you to expand on your reasoning against this voting method? What exactly is the nature of the objection, and which alternative voting systems would be preferable in that light? | |
| S Oct 2, 2013 at 1:41 | history | answered | Henry Cohn | CC BY-SA 3.0 | |
| S Oct 2, 2013 at 1:41 | history | made wiki | Post Made Community Wiki by Henry Cohn |