Re: [EM] question about Schulze example (A,B,M1,M2)
capologist wrote: See section 5 of my paper: Not quite what I'm looking for. That section describes a non-deterministic method for generating a complete linear order. I don't require a linear order. I'm OK with a partial ordering. I'm looking for a deterministic method for generating a picture (partial ordering) of how the voters, in aggregate, feel about the preferability of the available options. (What we're doing at this stage is more akin to a poll than an election.) It seems to me that the A(M1,M2)B ordering does not reflect the voters' preferences as well as the AM1M2B ordering. I'm open to the possibility that the Schulze method is the wrong tool for this purpose. I'm also open to the possibility that the Schulze method is the right tool for this purpose, and is serving that purpose effectively in this scenario. That would imply that, in some meaningful sense, A(M1,M2)B is at least as good or a better picture of the voters' preferences than AM1M2B. This is counterintuitive but perhaps it makes sense and I don't yet understand why. I think the latter is likely the case. M1 and M2 are beatpath tied. What's going on in this example is that there is a beatpath of strength at least 2 (using margins) from every candidate to every candidate. Since M1's pairwise win over M2 is not stronger than this value, it has no effect. Is this a case of a meaningful but weak signal being lost in noise? Or is the strength-2 cycle itself a meaningful signal that, for good if inscrutable reason, overrides the weak preference between the clones? The Schulze method works in a manner that's akin to shortest path (strongest beatpath). Now it might be that the shortest path to two different places from all other places being considered are the same, but if you consider every possible path, not just the shortest ones, it's possible to get to one of the places more quickly than the other. The analogy in Schulze is that even though the strongest beatpaths don't discriminate between M1 and M2, other non-strongest beatpaths might. Schulze, however, doesn't take these into consideration. There are two reasons for this. First, I think, is to make it more robust to noise and strategy. The second is that Schulze is intended to be somewhat of the closest method to minmax that we can have while also having Schwartz and independence of clones. Minmax also uses a similar best worst (strongest/shortest) strategy, and so scores M1 = M2. I imagine it would be possible to extend Schulze just as I have extended Minmax. My extended Minmax breaks ties (best worst defeat) by considering next-to-worst, then next-to-next-to-worst, and so on. The problem is that doing this with Schulze would involve finding not just the strongest beatpath, but the next-to-strongest, etc, down to the weakest; and if strongest beatpath is similar enough to shortest path, then weakest beatpath is similar to longest path, which is NP-hard as an optimization problem and NP-complete as a decision problem. So, to sum all of that up: Schulze ties M1 and M2 because it deliberately only considers the strongest complaints against putting candidates at a certain position. It does this to be similar to Minmax, which is like that (as far as I understand) to deter strategy. If you want to break ties, you could make an extended Schulze (but it could run for a very long time), or you could (for instance) break them in Ranked Pairs order. A Ranked Pairs tiebreak is fully deterministic. Sort the victories in order of magnitude, then if M1 M2 comes before M2 M1, set M1 above M2. It may feel hackish to transplant parts of Ranked Pairs into Schulze, however. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Proportional, Accountable, Local (PAL) representation: isn't this a big deal?
Jameson Quinn wrote: 2011/10/25 Kristofer Munsterhjelm km_el...@lavabit.com mailto:km_el...@lavabit.com Jameson Quinn wrote: * A multimember-district system helps with the above problems, but doesn't actually solve them. Who wants a system where ballots are only a little bit too complex, where you only sort of know who your representative is, and which is only mostly proportional? Multimember systems have been used in the US, on a local scale. The lack of such systems in the current day might just as well be due to that there is no modern day League of Proportional Representation such as the one whose efforts helped get STV into New York, than that multimember systems themselves are too complex. Fair enough. But note also that this was just the lesser of my two stated hurdles to MMP. STV is not mixed member proportional. As for the complexity issue, STV seems to work where it has been implemented. I agree that complexity will put a bound on how large each district can be, but as long as you keep below that size, it should work. If you have a district size of 5 members and 10 parties, that would give a seemingly unmanagable number of 50 candidates. However, voters can chunk by considering these candidates in party order. First they can consider do I like party A more than party B, then which of A's members do I prefer?. They do not have to rank all 50 members either, and few would. To the extent that the voters chunk in this manner, it seems to be personalized enough that the system doesn't degrade into party list (except in places where full ranking is enforced), yet it makes the burden easier to the point that ranked multimember voting does work. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] A structural fault in society owing to a design flaw in the electoral system
Dear Juho, Fred and others, I discovered something in history that enabled me to include formal equality in the thesis, along with electoral power. I post an expanded abstract/outline for critique. Can anyone see a weak point in the reasoning here? An individual vote in an election has no meaningful effect in the objective world, and no effect whatsoever on the official outcome of the election; whether the vote is cast or not, the outcome is the same regardless. Beneath this fact lies a structural fault that emerges here and there in society as a series of persistent discontinuities between facts and norms, or contents and forms. One rightly expects to be free because he lives in a democracy and has a vote; but the truth is, he has no political freedom at all. I trace the underlying cause of this fault to a technical design flaw in the electoral system wherein the elector is physically separated from the ballot. This separation removes the elector as voter (the active decider) from the social means and product of decision, thereby rendering him individually powerless. No electoral power exists in the vote itself, it exists purely in external communication networks; though the votes are brought together to make a result, the voters are not brought together *as such* to make a decision, consequently no valid decision can be extracted from the result. In the 1700s and 1800s, middle class society was able to partly overcome this flaw by engaging in politically animated practices of decision formation and expression that, even without the benefit of a concrete ballot, were nevetherless voter-like. This ad hoc practice of abstract voting enabled them to reconstitute electoral power within the flourishing communication networks of the day. As voting rights later expanded into the population, however, the franchise came to include more people who lacked the personal or social means to engage in abstract voting and reach decisions of their own. Their cumulative disengagement amounted to a power vacuum that coincided with the rise, after 1867, of the modern party system in Great Britain. The modernized Liberal and Conservative parties each responded by packaging its own ready-made decision, thus reducing the input of the elector to a choice of which package to consume. The resulting transfer of power from the weaker members of the electorate to the organized parties was the historical event that opened up the structural fault. It opened between the two formal components of political liberty, namely individual power and equality. These two components were torn apart for lack of any structural binding in society. Society is well equipped to handle the various forms of inter-personal or mass communication in which electoral power alone exists, but it lacks any concomitant support of equality. The ballot itself formalizes equality, but only internal to the electoral system; its structural strength cannot be realized unless it is externalized and personally bound to the elector. With that as a foundation, society could have provided electoral services on the basis of form rather than content; services in support of decision making as opposed to a one-size-fits-all consumption. Ordinary competition among service providers would then be sufficient to ensure that all electors regardless of personal and social means had access to their share of constitutional power and its associated opportunities. It was only ever a technical design flaw that precluded this development in the first place, and brought us instead to the present situation where the organized parties make the decisions and exercise the rightful power and political freedom that were intended for the citizens. [QCW] Please see also this revised figure, which is too large to post: http://zelea.com/var/db/repo/autonomy/raw-file/a44fa9a546c9/autonomy/a/fau/relations.png [REL] Causal relations among a formal failure of technical design (left) and actual failures in society (right). See descriptions in text of (a), (b), (s). The text itself is now divided into the following sections, which are still largely undrafted: [T] 1. The fact of an objectively meaningless vote 2. A structural fault in society 3. A design flaw in the electoral system 4. Abstract voting and the early public sphere 5. Franchise expansion, a power vacuum and the rise of the parties 6. A failure of structural support for equal opportunity Juho wrote: This [old abstract] was a bit too difficult to comment. The meaning of separation and its impacts are not clear. (No flaws identified, mostly opinions.) I think this new version is better at tying the formal separation (ballot from elector) to actual failures. Fred wrote: Where voting is by ballot, it is true that a voter who
Re: [EM] Proportional, Accountable, Local (PAL) representation: isn't this a big deal?
Interesting, but not relevant to what Kristofer had actually written. Finland uses a party-list voting system - Kristopher was writing about STV, and specifically about 5-member districts. James -Original Message- From: election-methods-boun...@lists.electorama.com [mailto:election-methods-boun...@lists.electorama.com] On Behalf Of Juho Laatu Sent: Saturday, October 29, 2011 5:11 PM To: EM Subject: Re: [EM] Proportional, Accountable,Local (PAL) representation: isn't this a big deal? On 29.10.2011, at 16.58, James Gilmour wrote: Kristofer Munsterhjelm Sent: Saturday, October 29, 2011 9:14 AM STV is not mixed member proportional. As for the complexity issue, STV seems to work where it has been implemented. I agree that complexity will put a bound on how large each district can be, but as long as you keep below that size, it should work. If you have a district size of 5 members and 10 parties, that would give a seemingly unmanagable number of 50 candidates. I think that is most unlikely. The only party that would likely nominate five candidates would be one that had reason to believe it could win at least four of the five seats in the multi-member district. Parties that might have an expectation of winning two seats would likely nominate only three candidates. Parties that expected to win only one seat would nominate at most two candidates, and based on our experience here in Scotland, many would nominate only one. So the total number of candidates in a 5-member district would almost certainly be far short of 50I think a total of 20 would be much more likely. Here's some data from last parliamentary elections in Finland. The largest multi-member district had 35 representatives and 405 candidates. All the large parties had 35 candidates. The largest party got 11 representatives. The two smallest multi-member districts had 6 representatives and 94 or 108 candidates. One of the parties grew from 5 representatives to 39 representatives. So it needed lots of candidates too in order to not run out of candidates in some districts. (see http://en.wikipedia.org/wiki/Finnish_parliamentary_election,_2011) If one has only one or two candidates more than the number of representatives that this party has or expects to get, then the decision on who will be elected will be mainly made by the party and not by the voters. Preliminaries could help a bit by allowing at least the party members to influence. If proportional results are counted separately at each district, then it would be good to have a large number of representatives per district to achieve accurate proportionality. In order to allow the voters to decide who will be elected there should be maybe twice as many candidates per each party as that party will get representatives. In that way no seats are safe. It is also good if there are such candidates that are not likely to be elected this time but that may gain popularity in these elections and become elected in the next elections. All this sums up to quite a large number of candidates. My favourite approach to implementing ranked style voting in this kind of environments would be to combine party affiliation and rankings somehow. The idea is that even a bullet vote or a short ranked vote would be counted for the party by default. If one looks this from the open list method point of view, this could mean just allowing the voter to rank few candidates instead of naming only one. Already ability to rank three candidates would make party internal proportionality in open list methods much better. Probably there is typically no very widespread need to rank candidates of different parties in this kind of elections, but it ok to support also this if the method and the requirement of simplicity of voting do allow that. From STV point of view the problem is how to allow better proportionality and voter decisions instead of party decisions in some nice way. Juho Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Proportional, Accountable, Local (PAL) representation: isn't this a big deal?
I just wanted to point out that actually one can come from open lists towards STV, and from STV towards a party based system with multiple candidates and end up pretty much at the same point. Juho On 29.10.2011, at 20.21, James Gilmour wrote: Interesting, but not relevant to what Kristofer had actually written. Finland uses a party-list voting system - Kristopher was writing about STV, and specifically about 5-member districts. James -Original Message- From: election-methods-boun...@lists.electorama.com [mailto:election-methods-boun...@lists.electorama.com] On Behalf Of Juho Laatu Sent: Saturday, October 29, 2011 5:11 PM To: EM Subject: Re: [EM] Proportional, Accountable,Local (PAL) representation: isn't this a big deal? On 29.10.2011, at 16.58, James Gilmour wrote: Kristofer Munsterhjelm Sent: Saturday, October 29, 2011 9:14 AM STV is not mixed member proportional. As for the complexity issue, STV seems to work where it has been implemented. I agree that complexity will put a bound on how large each district can be, but as long as you keep below that size, it should work. If you have a district size of 5 members and 10 parties, that would give a seemingly unmanagable number of 50 candidates. I think that is most unlikely. The only party that would likely nominate five candidates would be one that had reason to believe it could win at least four of the five seats in the multi-member district. Parties that might have an expectation of winning two seats would likely nominate only three candidates. Parties that expected to win only one seat would nominate at most two candidates, and based on our experience here in Scotland, many would nominate only one. So the total number of candidates in a 5-member district would almost certainly be far short of 50I think a total of 20 would be much more likely. Here's some data from last parliamentary elections in Finland. The largest multi-member district had 35 representatives and 405 candidates. All the large parties had 35 candidates. The largest party got 11 representatives. The two smallest multi-member districts had 6 representatives and 94 or 108 candidates. One of the parties grew from 5 representatives to 39 representatives. So it needed lots of candidates too in order to not run out of candidates in some districts. (see http://en.wikipedia.org/wiki/Finnish_parliamentary_election,_2011) If one has only one or two candidates more than the number of representatives that this party has or expects to get, then the decision on who will be elected will be mainly made by the party and not by the voters. Preliminaries could help a bit by allowing at least the party members to influence. If proportional results are counted separately at each district, then it would be good to have a large number of representatives per district to achieve accurate proportionality. In order to allow the voters to decide who will be elected there should be maybe twice as many candidates per each party as that party will get representatives. In that way no seats are safe. It is also good if there are such candidates that are not likely to be elected this time but that may gain popularity in these elections and become elected in the next elections. All this sums up to quite a large number of candidates. My favourite approach to implementing ranked style voting in this kind of environments would be to combine party affiliation and rankings somehow. The idea is that even a bullet vote or a short ranked vote would be counted for the party by default. If one looks this from the open list method point of view, this could mean just allowing the voter to rank few candidates instead of naming only one. Already ability to rank three candidates would make party internal proportionality in open list methods much better. Probably there is typically no very widespread need to rank candidates of different parties in this kind of elections, but it ok to support also this if the method and the requirement of simplicity of voting do allow that. From STV point of view the problem is how to allow better proportionality and voter decisions instead of party decisions in some nice way. Juho 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] question about Schulze example (A,B,M1,M2)
On Oct 29, 2011, at 12:29 AM, Kristofer Munsterhjelm wrote: you could (for instance) break them in Ranked Pairs order. A Ranked Pairs tiebreak is fully deterministic. Sort the victories in order of magnitude, then if M1 M2 comes before M2 M1, set M1 above M2. It may feel hackish to transplant parts of Ranked Pairs into Schulze, however. It may feel worse than hackish. I'm no expert in this field, but it is one I find interesting and visit from time to time. My first encounter with it was when I stumbled on a website advocating what was then called the Tideman method, before it was called Ranked Pairs and before the Schulze method was discovered. I had an email conversation with the author of that website during which I proposed several modifications that seemed to me to make sense. In each case he responded with examples demonstrating how my proposal failed important criteria and convincing me that it made the method worse, not better. From the experience I learned that these methods can have behaviors that are not obvious to me and that I should never use a method that hasn't been carefully vetted by people who understand the field much better than I do. You appear to be such a person. Would you say you have carefully vetted the suggestion you just made, or was it merely a thought off the top of your head? Election-Methods mailing list - see http://electorama.com/em for list info