Re: [EM] Question on RCV/IRV multi-seat method used in Minneapolis
First, I want to thank everyone who replied to my question. I'm buried in my usual unpaid work, or I'd be more conversant - trying to get three projects done in a very short time and no time for them all. I do appreciate all the answers and info. Date: Tue, 23 Sep 2008 16:01:46 -0400 From: Abd ul-Rahman Lomax [EMAIL PROTECTED] Actually, this is not the unfair part, and STV is actually quite fair, until it starts eliminating candidates. And that isn't the fractional vote part. When a candidate is eliminated, the votes transfered aren't divided (unless they were previously divided.) Yes. True when candidates are eliminated, but not true in the method used in Minneapolis and other locations when surplus votes above the threshold are transferred to other candidates. I may've misspoke, but that is what I was referring to. The *real* problem with IRV isn't monotonicity failure, but center squeeze, You call it center squeeze. I call it another case of a spoiler candidate whereby a spoiler is defined as any nonwinning candidate whose presence in the election contest changes who the ultimate winner of the contest is. Has anyone described the mathematical formulas for transferring excess votes above the threshold amounts ? I think I've figured it out. Few persons on this list seems to mind if an election method is incredibly complex and inconvenient to manually count or virtually impossible to manually audit in a way that the public can verify without doing a 100% hand count (and thus is difficult to ensure accurate and timely counts), but this method's description has been overly simplified and incompletely described where I've seen it posted. Admittedly, the publicly verifiable accuracy and integrity of the counts are my primary concern in evaluating any method, followed by fairness and ease of election administration and perhaps the other ten tenants of elections that I wrote about in a recent LWV, SLC article. Thanks for the assistance and answers to my questions. The topic of election methods is very intricate and complex and takes much time and talent to master. Cheers, Kathy Dopp Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Delegable proxy/cascade and killer apps
I imagine the biggest thing on offer (for N = 0 to 9) is the distant promise of what it's designed to do. We have to express that promise, and hold it out as worth reaching for (which it is!). Raph Frank wote: It has potential as an organising system. However, with only 10 people, it might be easier to just discuss the issue directly. Yes, if Nmax (population size) is low, then a voting medium is not needed. The purpose of the medium is support the growth of the discussion in the population as a whole. By providing structural handholds for agreement, it enables the discussion to climb in scale - like a rose bush on a garden trellis. I wonder if targetting a specific problem might help. Take something like a homeowner's association. That is large enough that not everyone bothers to get involved and small enough as a test system. Some have complained that the current system is not very effective and somewhat undemocratic. Your system could help people with very little time to participate. http://en.wikipedia.org/wiki/Homeowners_association There would probably be resistance to setting it up by existing directors. I'm not too familiar with private associations. My own test server is public, but that doesn't preclude other admins from setting up parallel servers for private associations. If anyone wants to use Votorola for that purpose, the code is free. (And I can probably help on the technical side.) There's a small side even to a big city server. It doesn't mean just big elections like Mayor. It also opens up the possibility of voting on local issues, such as neighbourhood improvement plans (like the park scenario in my other posts). That may be where the strongest roots will take hold - closest to home. -- Michael Allan Toronto, 647-436-4521 http://zelea.com/ Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Delegable proxy/cascade and killer apps
On 9/24/08, Michael Allan [EMAIL PROTECTED] wrote: Yes, if Nmax (population size) is low, then a voting medium is not needed. The purpose of the medium is support the growth of the discussion in the population as a whole. By providing structural handholds for agreement, it enables the discussion to climb in scale - like a rose bush on a garden trellis. Right, as these are hard to retrofit once power structures have been established. I'm not too familiar with private associations. My own test server is public, but that doesn't preclude other admins from setting up parallel servers for private associations. If anyone wants to use Votorola for that purpose, the code is free. (And I can probably help on the technical side.) I am not that familar with them either, was just trying to think of something that is small but still should have alot of the issues of a large deployment. There's a small side even to a big city server. It doesn't mean just big elections like Mayor. It also opens up the possibility of voting on local issues, such as neighbourhood improvement plans (like the park scenario in my other posts). That may be where the strongest roots will take hold - closest to home. True. There would be a wide range of scales in a city. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] (MA-1) A medium of communicative assent
On 9/24/08, Michael Allan [EMAIL PROTECTED] wrote: True, the translation barrier (from open to secret ballots) is another protection. It's partial. On its own, it cannot protect an open vote from purchase for its signalling value (like a paid endorsement, or a meeting stuffed with a paid audience). And it cannot protect norms, which are acted on by a different pathway. The fallback defence for these is recasting. Right, it is like currently, where control of the nomination process is powerful, even if it doesn't control the voters. You are thinking of using approval/range voting to provide an indicator of compromise *paths*? Interesting. It might be useful, especially for norms. Knowing that 2 candidate norms A and B *shared* assent (many approving of both) would reveal an opportunity to create a variant C that somehow combined the content of A and B. Assent might then shift to C. (It need not be a compromise document, technically speaking. If A and B are mutually compatible, then C might be purely an aggregate.) True. Also, lower level clients could approve multiple proxies. The end result might be that to many proposals are approved. It might be less useful for official elections, like for executives. The only way to combine executive candidates A and B would be for them to vote for each other as a team. Usually a team cannot occupy a single office. They occupy a power structure, usually with a clear chain of command. Well, candidate C might be the compromise. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Question on RCV/IRV multi-seat method used in Minneapolis
At 12:04 PM 9/23/2008, you wrote: Do you or does anyone know if this muti-seat IRV method that splits votes of voters to their second choice candidates after some winning candidates receive the threshold amount of votes, exhibits non-monotonicity or not like the normal IRV method does? If so, is there an example posted somewhere? Imagine a two-winner election using STV. The vote-splitting method doesn't matter because a scenario can be constructed where there is no vote splitting. If one candidate wins with an exact quota of first-place votes, those ballots are set aside and are not counted any more, because all those voters got their representative. Now the remaining ballots are used to elect a representative for the *rest* of the voters. And this is simply an IRV election. So you can use any IRV non-monotonicity example. If I'm correct, votes for the already-chosen candidate, found on these remaining ballots once elimination starts, will be disregarded and the next-lower choice awarded the votes, pending victory or elimination. STV is, as I've said many times, quite a good method, and it gets better the more candidates elected, *if* the electorate is well-informed and can coherently rank many candidates. Asset Voting finesses this problem, by essentially allowing voters to use their favorite candidate as a trusted proxy. Asset has other implications, though. It makes it possible for there to be a standing electoral college, consisting of public voters (all those who received votes in the secret ballot portion of the election), and this could even become a form of almost-direct democracy. But it starts as a simple fix for the ballot exhaustion problem that plagues IRV and to a lesser extent STV. The complexities of vote transfers are, in my opinion, justified by better representation. Very fair methods exist. I dislike random reassignment, but the mathematical methods, simplified to 1/1000 vote, seem fine to me. What does Cambridge, Massachusetts, do? (They've been using STV for a very long time.) (In Asset Voting, I'd use *exact* quotas, probably the Hare quota, -- votes/seats -- and exact fractional transfers, though they would presumably be rounded to some reasonable accuracy, sufficient to make a change of result from roundoff error very unlikely.) Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Question on RCV/IRV multi-seat method used in Minneapolis
At 03:49 PM 9/23/2008, Raph Frank [EMAIL PROTECTED] wrote: On Tue, Sep 23, 2008 at 5:59 PM, Kathy Dopp [EMAIL PROTECTED] wrote: It should be OK as long as the random selection is actually reasonably random. In theory. But Kathy Dopp is a voting security expert. They like to be able to recount elections and, if no errors were made, get the same results. While this can be done with pseudo-random sequences, I'm not sure I'd trust local election officials, who often have trouble counting plain ballots as it is, with the complexity. The vote transfers in STV, using strictly mathematical methods, can be done centrally. Deweighting is done with *ballots*, not with votes, as such. I.e., if a particular ballot has been part of a block of votes that elects a winner, it becomes deweighted accordingly. (If the exact quota was met, the deweighting is 100%.) Yes, it can get complicated. Usually, though, the number of winners is not huge, and deweighting only takes place when winners are created (and not the last one), so the number of possible combinations isn't huge. Asset Voting finesses the problem; I'd expect that most voters, even if the system remained STV, would simply vote for one. (I know a lot of people who think differently from me on this, at least at first. But if I don't trust a candidate to properly delegate authority, I probably shouldn't trust that candidate in the office, for they are going to make lots of decisions in office that I can't specify in advance, including many which delegate authority. Asset Voting would radically change the way we think about elections; current methods require us to make compromises with our trust, to choose to trust those whom we already perceive as being trusted by others. Asset Voting pokes strategic voting in the eye. Yet ... where is the chorus of support? When it comes to Range Voting and other methods, the bad strategic voting problems are trotted out ... but if we could eliminate it, make it essentially stupid and useless, nope, too radical. What if we could *trust* those we elect? The only way to get there is to vote only for people we trust! And current methods make this politically suicidal, they waste such fully-sincere votes, except rarely. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Question on RCV/IRV multi-seat method used in Minneapolis
On Sep 24, 2008, at 12:07 PM, Abd ul-Rahman Lomax wrote: In theory. But Kathy Dopp is a voting security expert. They like to be able to recount elections and, if no errors were made, get the same results. While this can be done with pseudo-random sequences, I'm not sure I'd trust local election officials, who often have trouble counting plain ballots as it is, with the complexity. The common approach, I think, is to sequentially number the ballots during the initial count so that the count is repeatable. It's worth noting that many (most?) STV counting methods are sensitive to ballot order to some extent. Meek/Warren is not, which is one of the attractions of those methods. It's also worth noting that, for all practical purposes, most if not all voting methods have a random element (or at least a means outside the actual counting rules) for breaking ties, and this must be taking into account during recounts. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] the 'who' and the 'what'
Good Afternoon, Michael This is in response to your message to me on September 8th. You describe what you have in mind via at least one level of abstraction and, for me, that adds a degree of difficulty. For example, and please forgive me obtuseness, I don't understand your closing paragraph: The point of my post is that we can actually do this today. It opens up an interesting question. In your own words: Would the voters be deciding on the 'who' and the 'what' in the form of candidates for the ballot, and norms for action? Or would they really (as McLuhan might suggest) be deciding on the whole electoral system? I believe you are referring to the mechanism on your site, but, even so, I don't understand the question. I have suggested that voters select nominees by meeting in triads and selecting one of their number to represent them. I'm unclear about how, exactly, you suggest that should or will occur. It's possible you have described these details on other threads and I missed them. If so, I apologize. I lack the time to digest all the material on this site, but do try to be thorough in any discussions I join. re: The elections are themselves an evaluative medium. Can that be true? When voting is based on media-disseminated obfuscation, deception and hyperbole, and when public susceptibility to such distortions are so well understood that spin doctors control the flow of information to the public, how can the resulting elections be evaluative of aught but the propagandists? Are the circumstances in which we find ourselves (in the United States) not proof of the fallacy of that point of view? re: The same communication channels that traffic in information about ordinary elections are also available for open elections. So voters have access to mailing lists and chat networks, blogs and broadcast media. They can use these media to share information and arguments about the candidates. At the risk of belaboring the point, these are precisely the means that foisted Weapons of Mass Destruction upon us and gave us our present crop of politicians. I'm surprised so few people recognize how the principles laid down by Pavlov, B. F. Skinner and a host of other behavioral scientists are used by our leaders (political and commercial) to milk us like cows. Mass communications is their tool and they are expert in its use. If we are to improve our electoral systems, one of our first concerns must be to find a candidate evaluation mechanism that goes deeper than the emotion-inspiring fluff we're fed by the media. Fred Election-Methods mailing list - see http://electorama.com/em for list info
[EM] Fwd: Question on RCV/IRV multi-seat method used in Minneapolis
(forwarding this) Hi, Did you intend to send your reply to the full list? I haven't replied to the list just in case. On Tue, Sep 23, 2008 at 10:38 PM, Abd ul-Rahman Lomax [EMAIL PROTECTED] wrote: At 03:49 PM 9/23/2008, you wrote: On Tue, Sep 23, 2008 at 5:59 PM, Kathy Dopp [EMAIL PROTECTED] wrote: It should be OK as long as the random selection is actually reasonably random. In theory. But Kathy Dopp is a voting security expert. They like to be able to recount elections and, if no errors were made, get the same results. While this can be done with pseudo-random sequences, I'm not sure I'd trust local election officials, who often have trouble counting plain ballots as it is, with the complexity. Officially, in Ireland, the ballot ordering must be preserved for the recount. Formally, the ballots are shuffled once at the start of the count and that is not repeated. I think recounts just count piles in place and confirm that they are possible ballots for that pile. The fun part is when they did the electronic voting in a few constituencies in Ireland, they had to use the same process as the hand count. Asset Voting finesses the problem; I'd expect that most voters, even if the system remained STV, would simply vote for one. (I know a lot of people who think differently from me on this, at least at first. But if I don't trust a candidate to properly delegate authority, I probably shouldn't trust that candidate in the office, for they are going to make lots of decisions in office that I can't specify in advance, including many which delegate authority. I think a combination ranked + asset is a reasonable compromise. It is possible to say, I like you but not your friends, especially to a politician. (Extra note) I am thinking of political parties here. Many voters cast non-party based voting in PR-STV elections, but candidates are likely to only transfer within the party. If Asset results in a very large number of candidates/electors, then this is probably less of a concern. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Question on RCV/IRV multi-seat method used in Minneapolis
There is no confidence in the US regarding EITHER the counting OR the maintenance of the rolls. Parties in power purge the rolls of voters known to be for party-not-in-power so regularly it's not even addressed by the courts any more, and NEGATIVE votes for candidates have been known to be certified as true by authorities who coincidentally are activists for other candidates in the election. There are several collection methods that preserve the secret ballot but allow auditability of election results. Until one of those are mandated in the US, no one in the free world should consider any US election (whether for mayor or president) honest and fair in any sense of the terms. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Raph Frank Sent: Wednesday, September 24, 2008 4:38 PM To: election-methods@lists.electorama.com Subject: Re: [EM] Question on RCV/IRV multi-seat method used in Minneapolis On Wed, Sep 24, 2008 at 9:25 PM, Kathy Dopp [EMAIL PROTECTED] wrote: This is similar in Ireland, confidence in the count is high. The electoral register on the other hand isn't trusted quite so much. Unfortunately in most US states, we have utterly SECRET ballot security procedures and no publicly verifiable ballot reconciliation, so that no one can know if the ballots have been tampered with, substituted, properly handled, or what. Secret ballot security measures are clearly a bad idea, especially if they are the only measures. Election-Methods mailing list - see http://electorama.com/em for list info Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Random and reproductible tie-breaks
In message [EMAIL PROTECTED] (on 24 September 2008 23:05:05 +), [EMAIL PROTECTED] (=?iso-8859-1?B?U3TpcGhhbmUgUm91aWxsb24=?=) wrote: for an anti-fraud purpose, the capacity to repeat the counting operation is a must. Hence I recommand to use a reproductible random procedure to break ties. This allows the use of different computers to reproduce the counting operation, while always obtaining the same result despite ties. Cryptographically secure hashing methods would appear to be the appropriate way to do this, using the ballots and/or some other agreed-upon information (e.g., the year of the election in a specified coding, the names of the candidates in a specified format...). I'm not sure whether it would be a good idea to set it up such that the tie-breaking couldn't be computed before the ballots came in (as would be possible if all the information needed for it was known in advance). -Allen -- Allen Smith, Ph.D.http://cesario.rutgers.edu/easmith/ September 11, 2001 A Day That Shall Live In Infamy II They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety. - Benjamin Franklin Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Random and reproductible tie-breaks
Hello Allen, simply using the number of ballots involved in the tie is enough. Compare its rest using euclidian divison by the number of involved candidates to the alphabetical rank of the candidates. Simple, effective and greatly equiprobable. It works for winner selection as for elemination rounds. Stéphane From: Allen Smith [EMAIL PROTECTED] To: [EMAIL PROTECTED] Subject: Re: [EM] Random and reproductible tie-breaks Date: Wed, 24 Sep 2008 20:37:03 -0400 In message [EMAIL PROTECTED] (on 24 September 2008 23:05:05 +), [EMAIL PROTECTED] (=?iso-8859-1?B?U3TpcGhhbmUgUm91aWxsb24=?=) wrote: for an anti-fraud purpose, the capacity to repeat the counting operation is a must. Hence I recommand to use a reproductible random procedure to break ties. This allows the use of different computers to reproduce the counting operation, while always obtaining the same result despite ties. Cryptographically secure hashing methods would appear to be the appropriate way to do this, using the ballots and/or some other agreed-upon information (e.g., the year of the election in a specified coding, the names of the candidates in a specified format...). I'm not sure whether it would be a good idea to set it up such that the tie-breaking couldn't be computed before the ballots came in (as would be possible if all the information needed for it was known in advance). -Allen -- Allen Smith, Ph.D.http://cesario.rutgers.edu/easmith/ September 11, 2001 A Day That Shall Live In Infamy II They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety. - Benjamin Franklin Election-Methods mailing list - see http://electorama.com/em for list info Election-Methods mailing list - see http://electorama.com/em for list info
