Re: [EM] MultiGroup voting method
On Apr 7, 2007, at 8:01 , Abd ul-Rahman Lomax wrote: Consider this: if you have a system designed for one vote, what do you do if the voter marks more than one? Typically, it will be considered an error and the vote is discarded. Note that depending on the style of ballots the risk (and opportunity) of voting several candidates may not exist. See e.g. http://www.vaalit.fi/17098.htm and http://www.vaalit.fi/35412.htm. It is easier to arrange the option to vote for multiple candidates in single-winner elections (e.g. Approval) than in multi-winner elections. I don't have any favourites at the moment on how MultiGroup could be enhanced in this direction. Asset style vote splitting is one option but it may add more complexity than it brings benefits (at least if the ballots are as described above). Once you are going to consider having overlapping districts, i.e., there can be more than one representative who represents a particular geographic location, with some representing the entire state, my question remains. Why be more complicated? My default setting comes from the multi-party tradition where it is typical that regions are quite large and several candidates are elected from each of them (PR). The MultiGroup method makes it possible for some candidates to indicate that they represent e.g. the western section of the region, or that they represent the western section of region B and eastern section of region A. One interesting scenario would be to allow any candidate to collect his/her votes from area of his/her preference as long its size (in number of citizens/voters) stays within agreed limits. The basic assumption here is that the society wants to force proportional regional representation and therefore no candidate is allowed to collect votes fro the whole country. MultiGroup could be used to relax the fixed (and possibly too rigid) region borders. Candidates representing border areas could also get a natural base (for them) of citizens to represent. In addition to this kind of mandatory regional representation rules also smaller voluntary regions/groups could be used. Complex reforms are pretty unlikely to be implemented. I don't think that the citation of tradition as a difference makes sense. Neither of these is traditional. For an open list system the delta to MultiGroup is just to add some attributes in the candidate list after each candidate name and enhance the counting rules a bit to cover this. The structure of the groups (e.g. parties, regions and their relationship) could deviate only a little from what the grouping has been before. If we really want to go outside of *political* tradition, we could go to delegable proxy (though this, in fact, simply brings in long- standing tradition in corporate governance); Asset, as I've been describing it here, remains with a peer legislature. I have no particular need to step out of the existing traditions of any country/organisation/society. In some cases radical changes may be needed to improve the system, but in other cases small enhancements could make the difference and put the evolution in a constructive/positive track. I will also note that I did look over Juho's proposal, and, beyond seeing similarities to Asset, I didn't understand how it works. That should be taken as a flaw. (Certainly how it works may have been explained, but that I didn't see it readily means that it has not been explained in a way to make it easy to follow. Part of the problem could be the complexity.) I had one example but I admit that it was not very detailed and I used a lot of abstraction, not fixing the s function, the criteria for optimal outcome and the calculation process to be used. This mail is already getting long, but I give one additional short description of how the calculation could be done. Each voter votes one candidate. Each candidate may belong to various groups. Each group will be guaranteed a proportional share of the seats (on could use e.g. largest reminder as the criterion). Check all possible outcomes of the election (this is a laborious task but the idea is simple). The outcome that implements the proportionality best for all groups and candidates will be selected as the final outcome (best = largest deviation compared first, then next in case of a tie etc.). I think this is quite simple, with the exception of the computational complexity of checking all the possible outcomes (= some heuristic approximate algorithm can be used to fix that). Another existing stream with connections to multiple interests is the possibility to give proxies to different persons on different topics. What does this have to do with Multigroup? No tight connection. Just that it addresses the question of how to better address multiple topics like use of nuclear power, education and employment within one election
Re: [EM] MultiGroup voting method
Juho Sent: 07 April 2007 06:53 easy way to vote - MultiGroup (vanilla version) uses just one bullet vote, STV-PR requires more - ease of voting is good if one wants to maintain wide involvement among the citizens and direct individual level decision making among the voters in public elections Any form of bullet vote gives the parties de facto control. Elections are for electors - or at least, they should be! James Gilmour election-methods mailing list - see http://electorama.com/em for list info
Re: [EM] Portuguese dictator should be the greatest portuguese?
Welcome to the Electorama list. -- Even without Chicago style voting (vote early, vote often) in this phone poll, the small response rate (about 1.5% of the population) and rumors about two polarizing front runners made the results especially sensitive to sample/participation bias. That effect would probably bias the results for just about any election method. -- Depending on the insightfulness of your friends, your estimates are probably a closer approximation to what a random sample poll would produce. -- To the extent that creating a participation bias resembles the behavior of honey bees, some might argue this poll was really a form of range voting. However, I think the honey bee example confuses political activism/campaigning with voting. -- The debates, sometimes heated and emotional, about which is the best election method have morphed into debates, sometimes heated and emotional, about which is the most important election criterion. Arguably, this represents some measure of progress. -- There is no best election method. Preferences for various election methods ultimately are a matter of personal preference, which may be situational. -- There is no best election method criteria. Preferences for various criteria ultimately are a matter of personal preferences, which may be situational. Criteria do sometimes highlight meaningful differences between various election methods, but there is also sometimes an element of inventing or reinterpreting criteria to prove one's favorite election method really is best. -- I've offered some critiques of the basis for Bayesian Regret as used in IEVS as an election method criteria / evaluation measure. Look at the some of my postings and related threads in the EM archives during the last month or two if you are interested. Likewise there are good critiques of using Condorcet winners as a election method criteria / evaluation measure. See the preceding item. -- IEVS evaluation of strategy comparisons is weak, in part because the strategies IEVS uses vary significantly in how optimum they are for various election methods. A meaningful comparison would also have to consider other factors as well such as what kind of information is available, how much cooperation there is within groups of individuals that prefer a given outcome, what the risks are of using a given strategy, and what are the costs and benefits, other than the election result, for using a given strategy. I'm not aware of any decent attempts to create a good framework for such comparisons. -- The contest using plurality voting was probably considered a great publicity / propaganda success from RTP's perspective. Any popular attention on which dead king, poet, explorer/colonialist (or here in the US, which emerging entertainer, celebrity wannabe dancer, or tropical island castaway) is/was best, is popular attention directed away from what the politicians are doing. In that regard, television in many modern societies has, I suppose, largely replaced religion. -- If it motivates even just a few people to become interested in and better informed about the available election methods, there may come some good from it after all. -- David Cary Now that's room service! Choose from over 150,000 hotels in 45,000 destinations on Yahoo! Travel to find your fit. http://farechase.yahoo.com/promo-generic-14795097 election-methods mailing list - see http://electorama.com/em for list info
Re: [EM] Portuguese dictator should be the greatest portuguese?
Welcome to the Electorama list. -- Even without Chicago style voting (vote early, vote often) in this phone poll, the small response rate (about 1.5% of the population) and rumors about two polarizing front runners made the results especially sensitive to sample/participation bias. That effect would probably bias the results for just about any election method. -- Depending on the insightfulness of your friends, your estimates are probably a closer approximation to what a random sample poll would produce. -- To the extent that creating a participation bias resembles the behavior of honey bees, some might argue this poll was really a form of range voting. However, I think the honey bee example confuses political activism/campaigning with voting. -- The debates, sometimes heated and emotional, about which is the best election method have morphed into debates, sometimes heated and emotional, about which is the most important election criterion. Arguably, this represents some measure of progress. -- There is no best election method. Preferences for various election methods ultimately are a matter of personal preference, which may be situational. -- There is no best election method criteria. Preferences for various criteria ultimately are a matter of personal preferences, which may be situational. Criteria do sometimes highlight meaningful differences between various election methods, but there is also sometimes an element of inventing or reinterpreting criteria to prove one's favorite election method really is best. -- I've offered some critiques of the basis for Bayesian Regret as used in IEVS as an election method criteria / evaluation measure. Look at the some of my postings and related threads in the EM archives during the last month or two if you are interested. Likewise there are good critiques of using Condorcet winners as a election method criteria / evaluation measure. See the preceding item. -- IEVS evaluation of strategy comparisons is weak, in part because the strategies IEVS uses vary significantly in how optimum they are for various election methods. A meaningful comparison would also have to consider other factors as well such as what kind of information is available, how much cooperation there is within groups of individuals that prefer a given outcome, what the risks are of using a given strategy, and what are the costs and benefits, other than the election result, for using a given strategy. I'm not aware of any decent attempts to create a good framework for such comparisons. -- The contest using plurality voting was probably considered a great publicity / propaganda success from RTP's perspective. Any popular attention on which dead king, poet, explorer/colonialist (or here in the US, which emerging entertainer, celebrity wannabe dancer, or tropical island castaway) is/was best, is popular attention directed away from what the politicians are doing. In that regard, television in many modern societies has, I suppose, largely replaced religion. -- If it motivates even just a few people to become interested in and better informed about the available election methods, there may come some good from it after all. -- David Cary Food fight? Enjoy some healthy debate in the Yahoo! Answers Food Drink QA. http://answers.yahoo.com/dir/?link=listsid=396545367 election-methods mailing list - see http://electorama.com/em for list info
[EM] Finding SociallyBest. Is it impossible?
Dear all, This might be a stupid question but I was wondering if SociallyBest exists at all, and if some day it will be found. I notice that the current approach to find a good voting method is proposing one on the first hand, and then comparing its performance with other methods based on Bayesian Regret, which, by the way, I dont know how to evaluate. My question for the mathematicians in this list, or perhaps this is a suggestion, is why not trying a different approach: to evolve a voting method, using some symbolic regression technique, that minimizes BR? Choosing the correct input variables might be a very dificult task, or not? Some of them could even be, besides the ballots data, Honfrac, NumVoters, UtilMeth and/or IgnoranceAmplitude. Perhaps some artificial inteligence tool, like neural networks or genetic algorithms, or a combination of both, could be used to search SociallyBest (zero BR), or at least get near it. If such formula is found, it could be truly complex or iloggical, something like a black box voting method, but mathematically very good. Dont you believe this approach could be tried? Just to give you a hint, are you aware of a technique called Gene Expression Programming? GEP was developed by a portuguese scientist and there is a software tool called GeneXproTools that implements it (see www.gepsoft.com). This symbolic regression tool allows a user defined fitness function, which could be BR, and there is a demo version for download from their website. Sorry if this ideia is totally ignorant. Regards, Ricardo Carvalho PS: Could someone please provide a VB routine to calculate BR? - Clix ADSL até 24 Mb: a partir de 29,90/mês A Internet mais rápida do mercado, agora com chamadas grátis e downloads ilimitados! Saiba mais em http://acesso.clix.pt/ election-methods mailing list - see http://electorama.com/em for list info
Re: [EM] Finding SociallyBest. Is it impossible?
Perhaps some artificial inteligence tool, like neural networks or genetic algorithms, or a combination of both, could be used to search SociallyBest (zero BR), or at least get near it. If such formula is found, it could be truly complex or iloggical, something like a black box voting method, but mathematically very good. Zero BR is impossible with strategic voters; that would mean electing the candidate that maximizes aggregate utility. But if that's what you're doing, then voters will be motivated to lie about their utility functions. It doesn't matter what sort of contortions you use in designing the method. With honest, perfectly introspective voters, you could just ask everyone to report their utility functions and sum them up. But such voters are a fantasy. The difficulty with evolving a voting method is that you don't know what strategic voting would look like. Maybe you could evolve the voting strategies too, but I expect you'd have pretty major issues with local optima. Peter de Blanc election-methods mailing list - see http://electorama.com/em for list info
Re: [EM] MultiGroup voting method
At 06:01 AM 4/7/2007, Juho wrote: it is an imposed system that the party names are on the ballot at all That could also be called information It is one particular kind of information, one which provides information about candidates affiliated with a party and no information about candidates not. Some ballots provide that the occupation of the candidate shall be on the ballot. Which if course allows the incumbent to state that he or she is the incumbent If you allow one kind of information, you favor candidates who look good to the voter in the light of that information. It introduces a bias to provide one or two bits of information, I've never seen more than that. The pieces of information I've desscribed -- party and occupation -- favor party affliates and incumbents. Big surprise! One difference is that in MultiGroup the declared associations to different groups are used in determining which candidates will be (proportionally!) elected. What this must mean is that, effectively, the voting is for groups rather than for candidates. which in my view is the exact opposite of what we need Note that to some extent grass always looks greener at the other side of the fence. Current political systems may not work optimally. But also future and alternative political systems are subject to corruption. The claim is made. The proof is actually lacking. Delegable proxy, well implemented in a society which has learned how to use it, would be highly corruption-resistant. Essentially, there aren't any critical nodes to target. The obvious targets are high-level proxies, but high-level proxies can lose their power in a flash if their clients smell a rat. So the high-level proxies, who are generally proxies for quite sophisticated clients, have to be able to convince their clients that the proposed action (which is actually the product of bribery of the proxy) is the best action. Now, if these arguments exist, the corruption isn't necessary! Continuous efforts are needed to keep the system working. Well, sure. But how *much* effort? It's possible that a system could be designed that would take little effort to maintain. That, indeed, is the point of DP, it distributes the communication and oversight load. Sometimes it is better to change an old system to a new one, but often it is also enough just to remove whatever rotten apples there are and find ways how to avoid such problems to emerge repeatedly in the future. More often, it is all blamed on the rotten apples and the remedy is limited to tossing them out, to be replaced by more rotting apples until the system changes, the stone will continue to roll back down the hill. However, the DP revolution doesn't have to change the old system. It supplements it, watches it, uses it. *It does not take changes in the laws to radically revolutionize politics. Nor does it take massive efforts and vast sums of money. What it does take is for enough people to realize the nature of the problem and what, in that light, actually becomes a very obvious solution We aren't using what we already know. election-methods mailing list - see http://electorama.com/em for list info
Re: [EM] Finding SociallyBest. Is it impossible?
Peter de Blanc wrote Perhaps some artificial inteligence tool, like neural networks or genetic algorithms, or a combination of both, could be used to search SociallyBest (zero BR), or at least get near it. If such formula is found, it could be truly complex or iloggical, something like a black box voting method, but mathematically very good. Actually, Isaac Asimov speculated along these lines back in his early 1950s short story Franchise, where the super-computer Multivac selected one voter, asked him a lot of seemingly irrelevant questions, and from the answers determined how everyone else would vote in every election. Pure fancy, but undoubtedly motivated by the then-new science of opinion polling. Zero BR is impossible with strategic voters; that would mean electing the candidate that maximizes aggregate utility. But if that's what you're doing, then voters will be motivated to lie about their utility functions. It doesn't matter what sort of contortions you use in designing the method. In the short story, and surely in practice, the AI engine knows enough about the voter that it would account for the voter's tendency to lie. The voter isn't asked what their utility function is (good thing, I haven't gotten anyone to give me a reasonable definition of utility function except as an abstraction for the complicated way I prioritize issues and my comprehension of the candidate's positions on the ones that are important to me) - the voter is asked questions designed to profile the voter. Based on those, the AI engine decides what the voter's utility function is. With honest, perfectly introspective voters, you could just ask everyone to report their utility functions and sum them up. But such voters are a fantasy. Utility functions are a fantasy. But it's true that asking voters to define theirs wouldn't work. Can you define yours? I can't even get a good definition for what that means. The difficulty with evolving a voting method is that you don't know what strategic voting would look like. Maybe you could evolve the voting strategies too, but I expect you'd have pretty major issues with local optima. This is only true if you're basing the hypothetical AI engine on the principle of summing individual utilities. The meta-considerations I mentioned above include that you have to start with some axioms as underpinnings. It might turn out that minimizing BR isn't the same as determining what is Socially Best. election-methods mailing list - see http://electorama.com/em for list info