Re: [EM] IRV's adequacy depends on a two-party system
Just the subject line on this is the most amusing thing I've read on this list in a while. Well said, sir! On Dec 2, 2011, at 2:19 PM, MIKE OSSIPOFF wrote: David Wetzel said: s for center-squeezing, that's not really a problem in the US as a whole... Third parties are too small and scattered. [endquote] Ok, so David is saying that IRV is adequate adequate only in a two-party system. Mike Ossipoff Election-Methods mailing list - see http://electorama.com/em for list info
[EM] advocacy: Approval is premature compromise
I know that Approval is technically better than a lot of things, and I think it's better than IRV, but I want to argue that it's not good enough and we shouldn't aim low or advocate it too strongly. I've always been personally unsatisfied with the prospect of filling out an Approval ballot. Sure I can say that either Al Gore or Ralph Nader would be fine choices for President, but I don't get to say which one I like better. I think this psychological aspect is important. In my mind it might drive me to misjudge my proper approval threshold, and I think I'd be likely to approve too few candidates and tend toward pick-one. I also today see Approval as fitting the pattern of premature compromise in politics. Afraid that we might not be able to get the awesome thing, we start off only trying for the mediocre thing. We could have real universal healthcare or Obama-Romney-care. We could try for a budget that makes sense, or we could have a budget half full of cruft and with tax tweaks that make no sense because someone whined for it. If we're going to do this, we should do it right. Go all the way. Go for the best thing possible. Isn't that one thing that frustrates us so much with the IRV advocates? They recognize that election method reform is important, but then they go all-in on a mediocre reform. Anyway, that's my random afternoon strategy opinion, I could be wrong. Brian Olson http://bolson.org/ Election-Methods mailing list - see http://electorama.com/em for list info
[EM] Election Method API Design
There's been some talk lately of publishing good reference implementations of all the election methods we talk about in various forms of description. This inspired me to write up my thoughts on programming election methods. It's in the body of the message below, and also posted as a Google Doc https://docs.google.com/document/d/1QZqBNVYJuNIIaFxE7IUB4nPZhodFHJu2JOPNy7oBiF8/edit?hl=en_US -- Some thoughts I’ve had over the years of writing election code for simulations or counting of real votes. I assume an object or context of some sort. Usually this works out to the two minimum basic functions: ElectionObject.vote( data ) ElectionObject.getResults() The vote() Function voteSN( map of strings to numbers ) voteNN( map of numbers to numbers ) voteA( array of numbers ) The simplest and most unambiguous vote function API is to take a mapping from strings (choice names) to numbers (rankings or ratings). The API should be unambiguous about whether the numbers are rankings (1 best, 2 ok, 3 lesser, etc) or ratings (high better). I usually name functions “voteRatings” or “voteRankings” to reinforce this. For efficiency of storage this often internally calls a voteNN function after mapping choice names to choice index numbers. Another benefit of this API is that partial ballots completion is very easy to represent. A choice not marked on the ballot gets no membership in the mapping passed to the vote() function. There might be A through F on the ballot, but my vote might only be {A: 1, C: 2, E: 3} Voting a string map also makes write-ins automatically represented. On the down side, I have had to rearrange some election method algorithms to be able to use this API and transparently accept write-in choices. Curious readers should read my Condorcet implementation which grows the NxN tally array as choices are encountered. Often the simplest vote() function to implement is one that simply takes an array of numbers. The fixed length array corresponds to a fixed list of choices. I have used this form of implementation in C code for simulation of thousands of elections per second, counting hundreds of thousands or millions of votes per second. Getting Results getWinner() getResults() textResults() htmlResults() htmlExplain() If code needs to do something based on the winner, then there should be code to return the winner(s). Mostly I’ve needed this in simulations where I need to count up statistics on who won and their characteristics, or to fill in a pixel in a Yee Election Space diagram. Displaying results for people to look at is simpler. Just present the available data in a reasonable way. There is usually a simple table of candidate names and a score or count of votes they got. This can be returned programattically as a map from names to numbers and rendered by other code. For my uses I’ve found it useful to have each election method know how to render its own output to ascii-art tables or HTML tables. Furthermore an ‘explain’ method is a good idea. Multi-round methods such as IRV should show the internal state on each round. Condorcet can show the NxN tally table. The getResults() function might do the final calculation on whatever votes have been submitted by vote() up to that point. I write it so that it caches the results inside the context object. Calls to vote() clear that cache. Repeated calls to getResults() and related methods use the cached version if available. Adavced Methods: Summability It’s possible to have a ‘summable’ method where votes could be added up into many groups and those group results further added together to get the final result. Condorcet, Approval, Plurality and others are this way. They don’t need to have the full set of ballot data in the end, just a summation will do. The method to achieve this would look like this: ElectionObject.addSubElection(ElectionObject sub) On the down side, it’s hard to implement this and get it right for some methods, and it’s not really needed. Any modern computer could hold all the votes for every person on earth. There is no election too big. So, it’s a neat trick, and mostly interesting to election algorithm nerds, but I probably won’t be implementing this more than I have so far … until the next time it would just be too cool and needs doing. Data Formats To go with my preferred representation of a vote as a mapping between choice names and numbers, there are a couple good ways of storing that kind of data. My favorite by far is URL encoded. nameA=number1nameB=number2nameC=number3 On the up side, it is unambiguous and very hard to misinterpret. (The only possible misinterpretation is whether the numbers are ratings or rankings.) On the down side, it repeats the choice name a great deal and wastes some space. This can be mitigated by any common compression software such as gzip/zlib, bzip2 or zip. They all do a very good job of compressing frequently repeated substrings. Another
Re: [EM] Election method simulator code
I counter-recommend git. I don't like it. If you like the new 'distributed version control' system style, I recommend Mercurial. code.google.com also supports mercurial. My own election simulator is also up on google code, also with subversion. It's kinda hidden inside my project for multi-language (C/Java/perl) election method implementation library. http://code.google.com/p/voteutil/ http://code.google.com/p/voteutil/source/browse/#svn%2Fsim_one_seat On May 6, 2011, at 8:29 AM, Jameson Quinn wrote: I recommend you put it up on GitHub. Git handles versioning and source control for you, and github is a good place for people who want to suggest code changes to do it directly, so it's easy for you to just accept or reject those suggestions. If you don't want to have to learn Git's command-line interface, there are a few gui tools: you can use git-cola for making checkins, and giggle or gitg for looking at the history of checkins. 2011/5/6 Kristofer Munsterhjelm km_el...@lavabit.com Quite some time ago, I rewrote and expanded the singlewinner part of my election method analysis program, mainly to add a cache to make X,,Y and X//Y methods very fast if results for base methods and sets X and Y had been calculated earlier -- and to only calculate the pairwise matrix one instead of 200 times if I were to find the results of 200 Condorcet methods. The last week or so, I've been cleaning up that code, and a version is up on Google Code at http://preview.tinyurl.com/5rd5krp . It's only tested on Linux, has some known bugs, and the actual structure isn't documented apart from comments, but there it is. I'll probably continue working on it now that I know how versioning works :-) If anyone has any questions or want to add to it, go ahead and reply! Election-Methods mailing list - see http://electorama.com/em for list info Election-Methods mailing list - see http://electorama.com/em for list info Election-Methods mailing list - see http://electorama.com/em for list info
[EM] WSJ article about AMPAS voting
http://online.wsj.com/article/SB123388752673155403.html Cites our own Warren Smith! Clearly we've been going about advocacy all wrong. Politics is boring, we should appeal to American's fascination with celebrities and sports. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Yee-Bolson Diagrams
specifically, my results are here: http://bolson.org/voting/sim_one_seat/ http://bolson.org/voting/sim_one_seat/www/ Ka Ping Yee did them first. I think I did them second and bigger and more. I haven't done any in a while, but my source is out there and there are other implementations as well. On Jan 6, 2011, at 9:26 AM, Leon Smith wrote: Forrest Simmons has mentioned Yee-Bolson Diagrams, and I am aware of (and admire) the work by Ka-Ping Yee, but as somebody not that familiar with the election theory literature, I don't understand the reference to Bolson or who he (or she) is. Could somebody give me a pointer? Best, Leon Election-Methods mailing list - see http://electorama.com/em for list info
[EM] A turd by any other name
Someone needs to tell Thomas Friedman that Alternative Voting (IRV) isn't all it's claimed to be. http://www.nytimes.com/2010/03/24/opinion/24friedman.html Also apparently Larry Diamond, a Stanford University democracy expert (as cited by Friedman), who sounds like probably a smart and nice guy who mostly specializes in growing new democracies around the world, and that's great, but for all the international cultural issues he's probably dealt with maybe we could help him out a little with this one little detail. If you want to get people started out right with the best methods we know of right now, IRV ain't it. I posted this to the NYT comments section. dunno if the moderators will accept it. Friedman and Larry Diamond need an adjustment on a point of election theory. Alternative Voting, also known as Instant Runoff Voting, is actually a pretty bad reform only barely better than the current system. Burlington VT enacted IRV for their mayoral elections but in 2009 on only the second time they used it the system got the wrong answer and elected the wrong person. A year later they repealed IRV. Virtual Round Robin elections, often known as Condorcet's Method, don't have that flaw and are just as easy or easier to implement than IRV. A few people have latched onto IRV and promoted it a lot, perhaps somewhat staking their reputations on it now, but really for the same amount of work we could have much better reforms. Brian Olson http://bolson.org/ Election-Methods mailing list - see http://electorama.com/em for list info
[EM] Simulating multiwinner goodness
There was a question on the list a while ago, and skimming to catch up I didn't see a resolution, about what the right way to measure multiwinner result goodness is. Here's a simple way to do it in simulator: Each voter has a preference [0.0 ... 1.0] for each candidate. Measuring the desirability of a winning set of candidates is simply summing up those preferences, but capping each voter at a maximum of 1.0 satisfaction. Unfortunately, this won't show proportionality. If 3/4 of the population have a 1.0 preference for a slate of 3/4 of the choices, we would measure electing one of them as being just as good as electing the whole set. So, we could apply the quota. If a candidate is elected by 3 times the quota, only apply 1/3 of each voter's preference for that candidate to their happiness sum. Now the huge coalition with their slate elected should each add up to about 1.0 happiness, and smaller coalitions should get theirs too. This is sounding a bit like an election method definition, and I expect that this definition of 'what is a good result' does pretty much imply a method of election. At worst, given ratings ballots that we can treat as the simulator preferences, for not too large a set of winning sets of candidates, get a fast computer and run all the combinatoric possibilities and elect the set with the highest measured sum happiness. Another thing we could measure in multiwinner elections (and possibly single winner) is the Gini inequality measure. If we have a result with both pretty high average happiness and low inequality, that's a good result. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Simulating multiwinner goodness
A couple years ago I moved from the California Democratic Party Machine to the Massachusetts Democratic Party Machine. I'm not sad when my party wins, I'm sad when they run boring stick-in-the-mud establishmentarian candidates. I'd love pressure from other parties to keep them honest, and that's what a lot of this whole election method reform thing is about. Anyway, an election method can't (directly) give me better choices, but just help me and the rest of society choose from the options available at the time. I posit that a better choice method will (eventually) encourage the availability of better choices. The current pick-one-primary and pick-one-general favors the boring old establishment too much. If I can safely vote for the obscure but awesome candidate as my first choice, and the safe establishment choice as second or third, I think we'll se more interesting little guys, and sometimes they'll win. But we knew all that. /advocacy On Mar 11, 2010, at 9:42 AM, Terry Bouricius wrote: But obviously, real world satisfaction with an election outcome is not so straight forward. I may favor a certain slate of candidates, but feel huge dissatisfaction if they all win, such that there is no opposition in the legislative body to keep them honest. This is what happened for many in British Columbia in 2001, when the Liberal Party won 77 out of 79 seats in the Provincial legislature. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Simulating multiwinner goodness
On Mar 11, 2010, at 11:29 AM, Jonathan Lundell wrote: As with any choice system based on cardinal utility, there end up being two problems that are not, I think, amenable to solution. One is the incomparability of individual utility measures from voter to voter (and here we're talking about utility deltas, since the utilities are normalized to max=1.0). The other is that, even if comparability were solved, we don't have a means of, in the individual case, determining what they are. Arrow made the same mistake. We can't compare interpersonal utility, but in practice we do. We set everyone's utility to One. One person one vote. That's how much you get. In particular, reported utility isn't very useful, since for the system to work, we need sincere utility, and a utility-based system provides every incentive to strategize. And, as Terry suggests, it's not clear what we *mean* by utility here. Happiness with what? The outcome of the individual election? The makeup of the resulting legislature? The legislation resulting from that legislature? Reported utility is vulnerable to all kinds of noise, imperfect reporting, imperfect introspection, and so on. And yet this can be simulated. We can make sim people who are perfectly knowable, add that noise, run the election, and see what happens both compared to the noisy utility and true utility. When I did this it turns out there are some methods less vulnerable to noise! (Condorcet better, IRV, with it's non-monotonic threshold swing regions is more vulnerable to noise.) And even if we could somehow measure the voter's ultimate happiness as a function of legislative outcome and come back in time and cast a vote, we don't have utilities for the counterfactual alternatives. However attractive it might be to fantasize about functions from cardinal utility to social choice, it comes down to an attempt to square a circle or invent a perpetual motion machine. The attemp might be fun, but we know a priori that it will fail. Are we talking about real people or sim people? I think we can make simulations and models that are useful. Lots of people keep trying, including me. Or are you sayng that we can't reasonably make sim people whose knowable sim qualities bear any useful resemblance to the real world? We're talking about all kinds of mathematical properties of election methods, why not various measures under stochastic test? What would be a good measure? Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Two simple alternative voting methods that are fairer than IRV/STV and lack most IRV/STV flaws
On Jan 13, 2010, at 8:06 PM, Kathy Dopp wrote: 1. A rank choice ballot method: Any number of candidates may be running for office and any number allowed to be ranked on the ballot. Voter ranks one candidate vote =1 Voter ranks two candidates, denominator is 1+2 = 3 votes are worth 2/3 and 1/3 for first and second ranked candidates Voter ranks three candidates, denominator is 1+2+3=6 votes are worth 3/6 and 2/6 and 1/6 for 1st, 2nd, and 3rd choice respectively Voter ranks four candidates, denominator is 1+2+3+4=10 votes are worth 4/10, 3/10, 2/10, and 1/10 for 1st, 2nd, and 3rd and 4th choice respectively ETC. Just follow the same pattern This sounds like a variation on Borda count, but with an incentive to vote on fewer candidates. With smaller and smaller votes as I give more information, I should vote for one _maybe_ two choices. Why would I want to give my favorite a 4/10 vote when I could give them a 2/3 vote or a 1.0 vote? This is the wrong incentive. Giving more information on the ballot should be encouraged. 2. A point system where a total number of points per voter per contest may be allocated by the voter to any of the candidates running for office: Two candidates running for office, give all voters 2+1=3 votes to cast. They may cast all three votes for one candidate or split the votes any way between the two. Three candidates running for office, give all voters 3+2+1=6 votes to cast. They may cast all six votes for one candidate or split the votes any way they like between the three. Four candidates running for office, give all voters 4+3+2+1=10 votes to cast. They may cast all ten votes for one candidate or split the votes any way they like between the four. Five candidates running for office, give all voters 5+4+3+2+1=15 votes to cast. They may cast all 15 votes for one candidate or split the votes any way they like. This is equivalent to any other normalized ratings ballot. People vote ratings, but they all have the same voting power, either by straight sum of ratings or by geometric distance or something. In any system where the voter has to allocate the points themselves, there will be nasty strategic thinking going on to try and allocate the points best. If I vote simply and honestly, allocating points in alignment with how I feel about candidates, points not in differential between the top two candidates are wasted. Kathy, in my investigations of election methods, I started with straight rating summation as optimal, but normalized ratings as more fair, but then ran into the wasted-vote problem and settled on Instant Runoff Normalized Ratings ( http://bolson.org/voting/methods.html#IRNR ). Over the course of rounds of counting it reallocates your vote based on your original ballot to always be optimally applied to the choices available. Never mind the Instant Runoff part of the name, by using ratings ballots and considering the whole ballot at once, it's much better than the simplistic IRV. It's much less non-monotonic than IRV, and gets better answers in my simulations. You can compare the relative non-monotonic areas in these election space plots: http://bolson.org/voting/sim_one_seat/www/ http://bolson.org/voting/sim_one_seat/www/4a_IRV.png http://bolson.org/voting/sim_one_seat/www/4a_IRNR.png http://bolson.org/voting/sim_one_seat/www/4a_Condorcet.png Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] just to let you know ...
On Jan 6, 2010, at 6:49 PM, Terry Bouricius wrote: Actually, the opposition to IRV in Burlington is predominantly focused on the complaint that the plurality leader in the initial tally ended up losing in the runoff tally. That's stupid enough to get me a bit angry. They see a problem with IRV results. Going back to pick-one voting is sticking their heads in the sand and denying to see the problem. Because we have the full data dump from the rankings ballots we can do analysis and figure out what happened and how IRV was wrong (and how pick-one would have been wrong), but if they take that away then we just won't know again. Problem hidden! I really hope the forces of stupid don't win. Good luck up there guys. If there was going to be a big public meeting, I might even be tempted to drive up from Boston. Even if I didn't get to contribute much I'd be curious to see just how these things play out amongst real Americans who aren election theory wonks. Brian Olson http://bolson.org/ Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Block group map of Texas
http://bolson.org/dist/TX/txmask.png http://bolson.org/dist/ME/memask.png (All of the states should be available by similar URLs by state postal code. Note the odd XX/xx capitalization.) On Nov 28, 2009, at 6:57 PM, Raph Frank wrote: I generated some pdf maps of Texas. They are pretty big as they include all the lines that are included in the census data. They would stress even the best of computers. However, it shows the kind of shapes that are included in the census system. County - 1MB (7MB expanded) http://www.electionsciencefoundation.org/temp_images/county.zip Tract - 4MB (32MB expanded) http://www.electionsciencefoundation.org/temp_images/tract.zip Block Groups - 7MB (60MB expanded) http://www.electionsciencefoundation.org/temp_images/tract.zip I also did a block one, but it comes to 420MB. 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] Proportional Representation from Ratings Ballots
On Nov 5, 2009, at 12:37 PM, Raph Frank wrote: Party A (65 voters): A1: 0.7 A2: 0.7 B: 0 Party B (35 voters) A1: 0 A2: 0 B: 1 Round 1 A1: 45.5 A2: 45.5 B: 35 All 3 have exceeded the quota. A1 or A2 will be elected and that will cause the 2nd unelected member of that party to increase his vote since they exceeded the quota. Thus after round 2, both A1 parties will be elected. Oh, that is a problem. It gets the right answer if I use L1 norm instead of L2. I think L2 norm is going to work better for single-seat IRNR but L1 norm better for multi-seat. L2 inflates the amount of vote that winds up getting applied to multiple choices. Warren Smith pointed out some pathological cases like: what if everyone voted (1,0,0,0,0) or (0,1,1,1,1), but I just can't get worked up over how if everyone voted weirdly they might get a weird answer. I'm working up a set of election space diagrams for STV, IRNRP and an artificial method that maximizes happiness where every voter gets a maximum of 1 unit of happiness (no point in giving any one voter too many of their favorite choices). What I want out of a PR election method: 1. Evaluate the whole ballot at once (not just first place vote like IRV/STV). This tends to find better solutions. 2. No wasted vote. Choices that don't win (disqualified in intermediate rounds) don't detract from a voter's ability to elect the best choice they can. Voting for a very popular choice leaves you some vote left for less preferred choices. (STV with Meek redistribution does this OK) 3. Poportionality. Colloquially: An N% constituency should get about N % of the seats. 4. Utility. Maximize the sum happiness of the population. 5. Fairness. No one or no group should be unnecessarily shut out. 6. Strategy resistance. For any voter or bloc of voters, honest voting maximizes expected utility. Brian Olson http://bolson.org/ Election-Methods mailing list - see http://electorama.com/em for list info
[EM] Explaining PR
Catching up from a couple weeks ago, I just wanted to add my short- short version of explaining Proportional Representation that usually gets a good response from people: A 20% group should get 20% of the seats. It's pretty easy for people to be agreeable to that. I think in general it might be easier for people to be agreeable to goals of election reform, and relatively few actually care to analyze the mechanics of how it works. If they are interested in the mechanics, they usually just want to understand it enough to feel they wouldn't be cheated by a mysterious system. Explaining it to the point of, oh yeah, that sounds reasonable is often enough. Depending on the audience, it might even be better to say that a 10% group should get 10% of the seats. Many US Green Party people would love that. Another 'goal' statement I like to use: I want to vote for candidates, not parties. I think a fair number of people have heard vague ideas about how in some places you vote for a party and the party fills in seats with people depending on how many they get allocated. I don't trust party machines even when it's my party, so I want to vote directly for candidates. STV allows this, and I like that (never mind its warts, we'll figure out better ways that also allow candidate voting). Anyway, those are my pithy two bits for the PR debate for now. Election-Methods mailing list - see http://electorama.com/em for list info
[EM] multiwinner election space plots
http://bolson.org/voting/sim_one_seat/20090810/ I think a few of these plots show Single Transferrable Vote behaving badly in the same ways IRV does, with discontinuities and irregular solution spaces. I also ran Condorcet and IRNR using combinatoric expansion. Combinatoric variants of single winner election methods adapt to multiwinner situations by enumerating all possible winning sets of the available choices and using a simulated voter's preferences on the choices in each set to determine a preference for each winner-set. Voting on the n-choose-k preferences for winner-sets then procedes as for a single-winner election. I think based on this I'm going to have to think more about making native multiwinner methods. Combinatoric expansion gets pretty expensive for large numbers of choices or seats to elect. I had been kinda resigned to STV being the state of the art in multiwinner methods, but we seriously ought to be able to do better. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] multiwinner election space plots
On Aug 13, 2009, at 9:18 AM, Kristofer Munsterhjelm wrote: Brian Olson wrote: http://bolson.org/voting/sim_one_seat/20090810/ I think a few of these plots show Single Transferrable Vote behaving badly in the same ways IRV does, with discontinuities and irregular solution spaces. I also ran Condorcet and IRNR using combinatoric expansion. Combinatoric variants of single winner election methods adapt to multiwinner situations by enumerating all possible winning sets of the available choices and using a simulated voter's preferences on the choices in each set to determine a preference for each winner- set. Voting on the n-choose-k preferences for winner-sets then procedes as for a single-winner election. How does the combinatorial expansion work? The way you describe it, it seems like it's general purpose - that you could combine it with any single-winner method. It's pretty general purpose but works well when there are ratings backing each voter. It's easy to derive a rating for a winner-set by just adding up the individual ratings. There would be more ties if there was an initial conversion from rankings to ratings, as 1st + 4th would be equal to 2nd + 3rd. Do you have the source for this program, as well? There is a public read-only subversion repository, check it out with: svn co http://voteutil.googlecode.com/svn/sim_one_seat or browse at http://code.google.com/p/voteutil/source/browse/sim_one_seat/ I think based on this I'm going to have to think more about making native multiwinner methods. Combinatoric expansion gets pretty expensive for large numbers of choices or seats to elect. I had been kinda resigned to STV being the state of the art in multiwinner methods, but we seriously ought to be able to do better. You could try implementing my DAC/DSC-based method (see http://www.mail-archive.com/election-methods@lists.electorama.com/msg04001.html ) or Quota-Preferential by Quotient (QPQ, seehttp://www.votingmatters.org.uk/ISSUE17/I17P1.PDF ), even if the latter is nonmonotonic (to my knowledge). Thanks for the links. Those both seem to be party-based proportional systems and I have so far been going (for partly ideological and partly technical reasons) for party-agnostic candidate-based systems. But perhaps I have misunderstood 'setwise highest'. I must have missed these the first times around and am now reading up and thinking about them more. It may also be that the construction of the voter preference profiles (Gaussian centered on a particular point) means that the ideal maps will look like Condorcet majoritarian elections. If so, they won't help distinguish proportional methods from disproportional ones, only show errors like clone problems. Proportionality might show itself somewhat like the distortions Borda counts show in single winner elections: http://bolson.org/voting/sim_one_seat/zoomout/4a_Condorcet.png http://bolson.org/voting/sim_one_seat/zoomout/4a_Borda.png http://bolson.org/voting/sim_one_seat/zoomout/4c_Condorcet.png http://bolson.org/voting/sim_one_seat/zoomout/4c_Borda.png (above from this page http://bolson.org/voting/sim_one_seat/zoomout/ ) The lines shift a bit and twist in weird ways, but have basically the same shape and none of the IRV discontinuities. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Redistricting, now with racial demographics
As this isn't something I really want it's going to be hard to get motivated to work it out. That said I think the way to go about it is to make unbiased districts by my current district, then pick one district with the highest proportion of the desired minority to elevate and adjust all the districts until that one has a majority of the desired minority. Repeat one district at a time until there are enough (some states require two or three I think). On Jul 16, 2009, at 6:46 PM, Raph Frank wrote: Are you considering updating the algorithm to include majority minority districts? This would potentially decrease the legal issues with using it for districting. Election-Methods mailing list - see http://electorama.com/em for list info
[EM] Redistricting, now with racial demographics
I've updated my redistricting site ( http://bolson.org/dist/ ) to include the racial breakdown of all current congressional districts (sometimes interesting by itself) and that of the compactness based districts I have come up with. If you want you can jump directly to http://bolson.org/dist/ XX where XX is any US state abbreviation. Brian Olson http://bolson.org/ Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] A range-voting poll you can vote in
Interesting UI. I'd thought about doing something like that but got discouraged when I tried to make it work on Internet Exploder. Had it for Firefox and Safari briefly though. My apps are also still out there, doing simultaneous Condorcet counting, histograms and other methods. http://betterpolls.com/ http://poll.appspot.com/ On Apr 18, 2009, at 11:34 AM, Warren Smith wrote: http://vote.superduperapps.com/polls/vote/e9a89Nte_v67cjH9 try it. Poll created 18 April 2009. -- Warren D. Smith http://RangeVoting.org -- add your endorsement (by clicking endorse as 1st step) and math.temple.edu/~wds/homepage/works.html 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] Burlington 2009 IRV election pathologies - updated web page
I'm late to the party, but I downloaded their vote data and ran it through my software and wrote up a page on it here: http://bolson.org/~bolson/2009/20090303_burlington_vt_mayor.html I think the most important point is the fundamental failure of IRV as itself, not any quibbling over a bad ballot here or there. Also, I _like_ this stuff and didn't manage to read through Warren's verbose pages on the topic, and I have tried to be more terse. On Mar 29, 2009, at 5:02 PM, Warren Smith wrote: Bouricius and Richie (FairVote) objected to the fact that many of the lies in their multiyear propaganda/lying campaign to mislead millions about IRV had been refuted by us by analysing IRV's pathologies in the Burlington 2009 mayoral election. (A lot of the FairVote lies are conveniently outlined in GREEN in our http://rangevoting.org/Burlington.html and http://rangevoting.org/BurlResponses.html pages; more are documented at, e.g, http://rangevoting.org/Irvtalk.html .) Although most of their new objections were as-pathetic and misleadling as their usuals, they did contribute one correct one. As a result, we have added the following update section to this page: http://rangevoting.org/BurlResponses.html#update -- Warren D. Smith http://RangeVoting.org -- add your endorsement (by clicking endorse as 1st step) and math.temple.edu/~wds/homepage/works.html Election-Methods mailing list - see http://electorama.com/em for list info Election-Methods mailing list - see http://electorama.com/em for list info
[EM] Voting space graphs, varying sigma
http://bolson.org/voting/sim_one_seat/20081203/ Tight population grouping at the left moving to widely spread on the right. Brian Olson http://bolson.org/ (Sent from my iPhone) Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Yee/B.Olson Diagrams (YBD's): the next step
Disorganized thoughts: Standard deviation could be considered to be interchangeable with the spacing between the choices. Wider spacing is equivalent to tighter standard deviation. I keep imagining a way to explain this as starting with a blank black space and colored dots representing the choices. Then visualize the population as a grey cloud of various intensity hovering over the choice space. The animation moves the population center to near a choice, fills in a pixel accordingly, jumps to near another choice, fills another pixel, and after a couple repetitions of jump and pixel fill the cloud jumps to the top left of the range and begins an accelerating scan over the space filling in a final image. I like the idea of scanning over a quadrant to fully show what result images are possible and how they change. In the end a small-multiple layout would probably make sense. Printed it might be an 8x8 array of 1 images. Hmm, I could bash this out in perl this afternoon. Most of my spare compute cycles are running on my redistricting project right now, but I could probably make a rough draft with tiny 100x100 graphs and run that pretty quick. I expect that showing the effect of standard deviation on IRV should be showable by a linear series varying standard deviation on an image that in some default case shows an interesting failure mode. I'm guessing that the size and shape of the defects will scale in a pretty regular way based on standard deviation. On Dec 1, 2008, at 7:17 PM, [EMAIL PROTECTED] wrote: One of the loose ends of Yee/B.Olson (YBD's)diagrams is the lack of completeness of exposition in even the three candidate case as exemplified by two natural questions that the skeptical observer might pose: 1. How do we know that the candidate configurations in the extant diagrams were not chosen by people biased against IRV and in favor of Condorcet methods? 2. What about the choice of standard deviation for the voter distributions? A survey showing how the shape of the candidate triangle affects the diagrams would answer the first question, and another survey that varies the standard deviations would answer the second question. Appropriate graphical summaries of the surveys and a good exposition of the whole would make an outstanding article of Scientific American caliber. First, a few words about question 2: The standard deviation of the normal distributions of voters has no effect whatsoever on the YBD of a Condorcet method. An infinitesimal standard deviation would yield exactly the same diagram as would a standard deviation with an infinitesimal reciprocal. But YBD's for IRV can depend heavily on the standard deviation of the voter distribution. As the standard deviation is made to approach zero, the IRV YBD will approach the Condorcet YBD, no matter the configuration of the candidates. Now a few words about question 1: We need a way of representing in one graph all of the possible shapes of the candidate triangle. One way to do that is to fix the two endpoints of the side of largest length, and let the third vertex vary enough to cover all of the possibilities of shape. Suppose we put the endpoints of the largest side at (0, 1) and (0, -1) on the x-axis of a rectangular coordinate system, so that the largest side has length 2. Then without loss in generality we can restrict the third vertex to the first quadrant. And since its distance to the vertex (0,-1) is no more than two, we restrict further to the part of the first quadrant that is inside a circle of radius 2 centered at (0,-1). For graphical results, each point of that region is colored according to the type and extend of pathology associated with a candidate configuration of that shape. Some of the pathologies are these, in order of increasing seriousness: (1) a non-convex win region. (2) a win region that is not star-like with respect to its candidate. (3) a win region that is not path connected or one with a hole in it. (4) a win region that doesn't contain its candidate. (5) an empty win region. Of course this shape based graph will have no regions of pathology when the standard deviation is sufficiently small, but for large standard deviation we expect more than half of the graph to represent one degree or another of pathology. Yet another graph that relates the percentage of pathology in the shape graph as a function of standard deviation would complete the picture. In fact, there could be several such graphs, one for each kind of pathology. There is potential for a great Scientific American article here, not to mention a project for a master's degree. Forest 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] Why I think IRV isn't a serious alternative
On Nov 26, 2008, at 5:53 AM, Greg wrote: Greg, you didn't actually say that IRV is good, you just said that it's unlikely to be bad. Huh? One reason I think it's good in part because it's very likely to elect elect the Condorcet candidate, if that's what you mean by unlikely to be bad. Some other reasons I think it's good is that it resists strategic voting, allows third parties to participate, and paves the way for PR. And you get all those other good qualities in just about every other election method past 'pick one'. And I should have flipped that around, 'unlikely to be bad' means that there's a definite chance that it will be bad. Why bother with something that's unlikely to be bad when we can just as easily get something without that badness? You can't get rid of badness. Every system is imperfect. IRV is non-monotonic; Condorcet is susceptible to burial. So we're left to balance the relative pros and cons. So, you discount and ignore the possibility that IRV will make systemically the wrong choice and behave chaotically, and I discount and neglect the possibility that some people will cast crazy rankings ballots. Obviously I still think I'm making the more rational evaluation. Oh, and actually it _is_ likely to be bad. See that first graph? See how over thousands of simulated elections it gets lower social satisfaction? Brian, you're graphs are computer-generated elections that you made up. They aren't actual elections that took place in practice, which show a high unlikelihood of being bad. When your theory is a poor predictor of the data, it's time to change the theory, not insist the data must be different from what they are. Given the substantial lack of data (pretty little real world rankings ballot data available), I think the simulations are still valid and interesting. The simulations explore a specific and small portion of the problem space in detail. I'm looking at races of N choices which are similarly valued by all the voters. It's a tight race. Actual elections haven't been that tight. But tight races are the interesting ones. When it's crunch time, those are the ones that matter. Almost any method can correctly determine the winner of a race that isn't tight. So, IRV has demonstrated in the real world that it can solve easy problems. So what? Why wait until it gets the wrong answer in a real election to admit that IRV can get the wrong answer? In matters of public safety that would be called a 'tombstone mentality'. Election-Methods mailing list - see http://electorama.com/em for list info
[EM] Why I think IRV isn't a serious alternative
There's been a lot of discussion lately started by people who advocate IRV. I'm mystified. Really? You really think IRV is a good system? I've spent so long considering it to be pretty much junk that I really am confused by that position. Here's my summary of why I think IRV is junk. (from http://bolson.org/voting/irv/ ) First, a concession Yes, IRV is better than nothing. It's better than tired old pick-one voting. But that's about all it's better than... IRV Gets Worse Results The graph below shows an assortment of election methods run in simulated elections of 1 voters and various number of candidates to choose from. Increased happiness is on the virtical axis. In general, as there are more choices, there's a better chance of there being some choice that makes more people happy, and so overall happiness rises. If you can only cast one vote this starts to fail above 7 choices as the populace gets fractured voting for their favorites. At the bottom is a control test experiment of picking a choice at random, which on average makes half the people happy, and half unhappy, for a total of zero. Third from the bottom is Instant Runoff Voting. IRV doesn't get very good overall results because it only considers a small part of the votes at once, the top ranked choice. Because of this it can miss broadly supported compromise candidates. IRV doesn't benefit as much from having a wide selection of candidates to choose from. And as the next section shows, it can flat out get the wrong answer. IRV Gets Chaotic The diagrams below illustrate the decision reached by two election methods depending on where the center of public opinion lies. If the center of public opinion is closest to a candidate (diamonds) then the election should go to that candidate and the diagram gets colored to match. The plot of IRV results shows that IRV is capable of completely missing giving the election to the blue candidate even when the population is right there. In other regions the result is chaotic. IRV has this irregular behavior because of how it transferrs lots of votes and disqualifies candidates. Subtle differences in the order of disqualification of candidates and the transfer of votes to other candidates can wind up swinging the election to substatntially different outcomes. Instant Runoff Vote Virtual Round Robin Election (Condorcet's Method) more election diagrams IRV Doesn't Scale Up IRV requires all the ballots to be in one place at one time for counting (when someone's top active choice is disqualified, you have to go back and check their ballot to see how they would vote for their next most preferred). Or, for computer counting, the data from all the ballots has to be gathered in one computer. Hand counting techniques require physically moving piles of ballots around, and in a large election this could wind up requiring a fork lift. VRR/Condorcet can be summed up from intermediate results, counted at precincts or counted by many people hand counting ballots. Hand counting techniques just need a sheet of paper to make tally marks on. FUD: Other Methods Hurt Your Top Choice The simplistic method of Instant Runoff Voting, always counting only your top ranked choice (of candidates not eliminated in the rounds so far), is philosophically attractive to some people, but may not actually best serve your interests. One way of thinking about how IRV works is that if you were to assign numeric ratings to your choices, you might give a 1 to your fourth choice, 10 to third, 100 to second and 1000 to your favorite. If this huge difference in preference between choices accurately represents how you feel, IRV may represent you relatively well (but it can still make the generally wrong choice, as shown above). Other methods tend to have a more uniform effective distribution of preference between the choices. A second or third choice obviously isn't as preferred as the first choice, but out of several candidates, second or third choice is probably still good or ok. IRV won't consider these lower ranked choices at all, and may have prematurely disqualified them if they didn't get enough first-choice vote. There's a more general rumor applied to IRV and other rankings methods in general that says you should only vote for a top choice, or a top few, and not give any ranking to lower preference choices so that they will not be aided in any way. For any decent election method, and even IRV, your lower ranking choices will only be relevant if your higher ranking choices are losing anyway. Sometimes, that's just Democracy, and you don't get what you want. A fully ranked ballot at least gives you some say in which of the other ones gets elected. You might not get what you want, but you might get someone less bad than worst. FUD: Other Methods Dilute Your Vote Similar to the
Re: [EM] Why I think IRV isn't a serious alternative
Greg, you didn't actually say that IRV is good, you just said that it's unlikely to be bad. Why bother with something that's unlikely to be bad when we can just as easily get something without that badness? Oh, and actually it _is_ likely to be bad. See that first graph? See how over thousands of simulated elections it gets lower social satisfaction? On Nov 25, 2008, at 11:52 AM, Greg wrote: I will believe that when I'm presented with a non-negligible number of actual IRV elections for public office that failed to elect the right winner. And for starters, you get to define what right is. Preferably something of the form: in Election X, IRV elected candidate Y but candidate Z was the right winner, because of [insert your criteria and evidence here]. The more such cases you have, the more convincing your argument. I've studied every IRV election for public office ever held in the United States, most of which have their full ranking data publicly available, and every single time IRV elected the Condorcet winner, something I consider to be a good, though not perfect, rule of thumb for determining the right winner. When you present a case in which IRV did not elect the right winner, maybe I'll agree or maybe I'll dispute your criteria, but at least then we'd be off the blackboard and into the world of real elections. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] IRNR question
Hmm, only kick out the losingest loser. I kinda think there would still be discontinuities, but it might be better. Probably worth trying. Now I just need to code that up and run the diagram code. Dunno when I'll actually get around to that. Has anyone checked what happens to regular IRV under such a system? On Oct 19, 2008, at 2:35 AM, Greg Nisbet wrote: Would Brian's IRNR benefit from an addditional level of recursion? The current way to eject candidates is to compare range scores, what if you modify that slightly? Instead of kicking out the person with the lowest range score you replace that with: Kick out the person with the highest range score, shift the ratings and do the same thing again. You are left with one candidate. Kick this candidate out from the main system and repeat the above step. Just as a broader question, do methods such as IRV, Nanson, Baldwin, IRNR generally perform better or worse as additional levels of recursion are added? 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] Populism and Voting Theory
On Oct 17, 2008, at 9:44 AM, Raph Frank wrote: Anyway, you would rank PR-STV behind single winner election methods? As a priority of things to do? Yeah kinda. It's substantially a separate issue. There will be single winner elections (mayor, governor, president, other one-off seats), and there will be multi- member bodies and some of those should be converted to a PR system, and for the time being getting better single winner elections could apply to all those districted elections. So I think getting ranking/ ratings ballots on single winner votes is the single biggest change we could make to the electoral system. But hey, follow your passion. There are plenty of good things to do and we should do them all and I think we're most effective when we're working on what we personally care most about and in coalition with the right allies even if they're focusing on different aspects of the movement. CPO-STV (or maybe Schulze-STV) are obvious improvements, but with big costs in complexity. I do think that vote management is a weakness of PR-STV (I wonder if Schulze STV would stop parties bothering to try). Also, the district sizes need to be reasonable (say 5+). In Ireland, there are 3.86 seats per constituency on average, which I think is to low. Oops, I may have written imprecisely. I meant PR-STV to mean the general philosophy of having Proportional Representation governing bodies, likely elected by a variation on STV. Also, if you could make one change, would you implement IRNR or redistricting reform? Unfortunately, with extreme gerrymandering, I think most methods would still elect a member of one of the two parties. I'm still going for changing single-winner election methods as the biggest change, and likely biggest bang-per-buck we can get out of changes to work on. Brian Olson http://bolson.org/ Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Fixing Range Voting
On Oct 15, 2008, at 1:59 PM, Peter Barath wrote: I'm not sure I would vote honestly in such circumstance. Let my honest rangings be: 100 percent for my favourite but almost chanceless Robin Hood 20 percent for the frontrunner Cinderella 0 percent for the other frontrunner Ugly Duckling I think I would vote: 100 Robin Hood; 99 Cinderella; 0 Ugly Duckling If I'm really sure that the race decides between Cinderella and Ugly Duckling, why care too much for poor Robin Hood? And what, if I'm not really sure, because that's the situation which multi-candidate voting is really about? If I lower Cinderella's 99 to her honest 20, I make Robin Hood a little bit more hopeful not to drop first. But more hopeful against whom? Cinderella, of course, because I didn't change Robin and Ugly's obvious rangings. So I made more probable a situation in which more than 50 percent is the probability that the worst candidate wins. This is a doubtful advantage. On the other side, there is the effect that by rising Cinderella's points from the honest 20 to 99 I made more probable the similarly unlikely but positively desirable effect of Ugly dropping first instead of her. So, which does have more weigh? The doubtful little hope for Robin Hood, or the clear little hope against Ugly Duckling? I think the latter. Maybe at some point, let's say Cinderella's 5 percent, I like Robin so much more that I chose the first one. In that case I probably would vote 100-1-0 These voting are not the honest although by one percent honer than the simple Approval voting. But I would be open for persuasion. If you vote (100,20,0), (100,99,0) or (100,1,0), if your 100 hero loses in the first round, your vote in the second round is (x,100,0). So, what are the various consequences in the first round vote, in case it makes a difference there? I think the normalization comes into why you want to vote differently. (100,20,0) = (98.1,19.6,0) (100,99,0) = (71.1,70.4,0) (100,1,0) = (99.995,0.5,0) I think the tradeoff is that in a many-candidate race your lower preferences might contribute to runoff-disqualification order. You can put the vast majority of your vote on your favorite, and that's ok and your vote will get transferred to the remaining candidates if you don't get that favorite, but your lower rated choices might still be affecting which choices are disqualified or remaining at that time. The 100,99 vote looks tempting because it normalizes to a lot of absolute value, but that does come at the price of losing some weight on your favorite and making your 2nd choice a bunch more likely to win. I think it's this tradeoff that will squeeze people towards voting honest ratings. I could see honest voting want any of these three votes. Wanting A or B vastly more than C, wanting A vastly more than B or C, or some more gradual falloff. Does IRNR not do the right thing for those three voters? Election-Methods mailing list - see http://electorama.com/em for list info
[EM] Fixing Range Voting
With all the talk about Range Voting and its plusses and minuses, I wanted to inject this back into the mix. Once upon a time, I designed an election method to fix the strategy problem with Range Voting. The strategy problem: You shouldn't cast a ballot with your honest ratings, you should maximize them along Approval strategy lines. It also fixes the counting problem of how if someone does cast votes throughout the range, they might have done better in the end by different values. The method I call Instant Runoff Normalized Ratings (IRNR): 1. Collect ratings ballots 2. Normalize each ballot so that each has an equal magnitude 3. Sum up normalized ballots 4. If there are more than two choices, drop the one with the smallest sum. If there are two choices remaining, one is the winner. 5. Re-normalize from original ballot values but as if dropped choices weren't there 6. Go to 3 I think it gets very near to a utilitarian ideal solution ( http://bolson.org/voting/twographs.html ) and encourages people to vote honestly and uses those honest votes to the best possible effect. Its Instant Runoff nature does have some drawbacks. It is not summable by parts and requires all the data to be collected in one place. It also has some small discontinuities in the Ka-Ping Yee diagrams: http://bolson.org/voting/sim_one_seat/www/4c_IRNR.png But at least it's not as bad as IRV: http://bolson.org/voting/sim_one_seat/www/4c_IRV.png I have some ideas about smoothing out the discontinuity, but haven't gotten around to trying it yet. I think the key is to make the process more continuous and take smaller steps. Don't disqualify a choice all at once, but over several steps. Blend out the losing choices, blend out the nasty jumps in the decision process. Needs to be experimentally (in simulator) checked, though. Brian Olson http://bolson.org/ Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Fixing Range Voting
On Oct 14, 2008, at 12:11 PM, Raph Frank wrote: On Tue, Oct 14, 2008 at 2:04 PM, Brian Olson [EMAIL PROTECTED] wrote: Once upon a time, I designed an election method to fix the strategy problem with Range Voting. The method I call Instant Runoff Normalized Ratings (IRNR): 1. Collect ratings ballots 2. Normalize each ballot so that each has an equal magnitude Magnitude = max rating - min rating or Magnitude = sum of ratings (negative ratings allowed) ? I mean the geometric sense. For ratings a,b,c,etc., sqrt(a*a + b*b + c*c ...) 3. Sum up normalized ballots 4. If there are more than two choices, drop the one with the smallest sum. If there are two choices remaining, one is the winner. 5. Re-normalize from original ballot values but as if dropped choices weren't there 6. Go to 3 I think it gets very near to a utilitarian ideal solution ( http://bolson.org/voting/twographs.html ) It id heard to determine which plot refers to which method. In a sense, part of the result is that there's a pretty tight pack of similar (good) results and a few outliers (IRV, pick-one). and encourages people to vote honestly and uses those honest votes to the best possible effect. Maybe, I can't see any obvious strategy and it seems to protect people from casting weak votes. Its Instant Runoff nature does have some drawbacks. It is not summable by parts and requires all the data to be collected in one place. It also has some small discontinuities in the Ka-Ping Yee diagrams: http://bolson.org/voting/sim_one_seat/www/4c_IRNR.png But at least it's not as bad as IRV: http://bolson.org/voting/sim_one_seat/www/4c_IRV.png I have some ideas about smoothing out the discontinuity, but haven't gotten around to trying it yet. I think the key is to make the process more continuous and take smaller steps. Don't disqualify a choice all at once, but over several steps. Blend out the losing choices, blend out the nasty jumps in the decision process. Needs to be experimentally (in simulator) checked, though. I am not so sure candidates need to actually be eliminated. Candidates who are out of the running would still be rated by each voter. However, they would not be used to determine the truncation window used. Ofc, that could mean that the method doesn't converge. This is especially true if there is a condorcet cycle. For example, if a voter votes A: 10 B: 3 C: 0 initially, the vote will just be rescaled to give maximum window = (0,10) A: 1.0 B: 0.3 C: 0.0 Each candidate would be assigned a score based on the result of the first round, then the new window needs to be worked out 1.0 = in the running 0.0 = eliminated For each ballot. 1) work out weighted mean using the scores. - This will be the centre of the window 2) for each candidate work out d(candidate) = (score)*(distance from mean) 3) determine the highest distance 4) Set window to window( mean - dmax, mean + dman ) In the above example score(A) = 1 score(B) = 1 score(C) = 0.5 mean = (1*10 + 1*3 + 0.5*0)/2.5 = 5.2 distance(A) = 1*4.8 = 4.8 distance(B) = 1*2.2 = 2.2 distance(C) = 0.5*5.2 = 2.6 dmax = 4.8 window = (5.2-4.8 , 5.2+4.8) = (0.4, 10) The ballot would be rescaled as A: 1.0 B: 0.27 C: 0 If only 2 candidates remain, then it will set the window as max and min of those candidates. I think that's pretty similar to what I'd planned to implement. I'm still expecting some tinkering will be needed to get it to do solutions with negligible instability. Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] Why We Shouldn't Count Votes with Machines
On Oct 6, 2008, at 11:30 AM, AllAbout Voting wrote: So I will ask a pair of constructive questions: 1. Can Condorcet voting be compatible with precinct level optical scan systems? (which many election integrity advocates consider to be pretty good) Yes. 2. Can Condorcet voting be compatible with end-to-end verifiable election integrity systems such as punchscan, 3-ballot, etc...? Aside from the NxNx3 adaption of 3-ballot to condorcet information, I think it was suggested on this list a while ago that 3-ballot can be adapted to 0-100 range voting by scaling up its three ballots of 0-1 voting and requiring sums of 100-200 for a valid vote instead of sums of 1 or 2. If that sort of system was used, rankings for condorcet counting could be extracted from the ratings votes, or a more advanced ratings-aware system could be used. Actually, that sounds pretty messy. NxNx3 is probably better. Most of these methods require automatic ballot construction or specially clueful voters. I'd expect 99% of voters to never bother verifying that the election was actually done right if they had the certificates with which to check it. I think probably the best defense against electoral malfeasance is probably through the political and legal processes, and through the vigilance of the citizenry. We'll never make a system so mathematically perfect that we don't still need those other things. Brian Olson http://bolson.org/ Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] A computationally feasible method (algorithmic redistricting)
On Sep 1, 2008, at 7:28 AM, Kristofer Munsterhjelm wrote: A simpler algorithm, though one that doesn't give any worst-case bounds, is Lloyd's algorithm. Start with random center points, then calculate the centroids for the current Voronoi cells. Move the center points towards the centroid (shifting the Voronoi cell somewhat) and repeat. Do so until they settle. This can get stuck in a local optimum. I have implemented essentially that, and it pretty quickly gets pretty good results as measured by the distance per person to land-area- center test. I added one thing to deal with variations in population density. Each district center also has an associated weight. Each 'voronoi cell'/district contains the people with a weighted distance closest to that district center. So, a district center in a high population density area doesn't have to reach very far to gather its allotment of people. There are some states where this doesn't work very well (CA, TX, NV, and a couple others) and a more exhaustive solving method is needed. I think one of the weaknesses of the voronoi method is that population is not continuous, but in discrete census blocks of dozens to thousands of people. The voronoi method can't necessarily tweak to swap a block here and a block there to get within a reasonable margin of equal-population. I think it's worth repeating that I'd prefer to write the law with a way of measuring what a good district mapping is, rather than writing the law with a specific procedure for creating the district mapping. Specific procedures get really complex, and I don't think in this case that a fair procedure ensures a fair outcome. I used the voronoi technique as just one possible way of writing a solver to get the desired result. It could even wind up that the best way to solve for the desired goals is to have a computer do 99% of the work and have a human come back and clean up the edges. Since it's subject to the same scoring in the end, the human couldn't deviate too far from the ideal mapping. Brian Olson http://bolson.org/ Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] A computationally feasible method (algorithmic redistricting)
On Sep 2, 2008, at 6:00 PM, Kristofer Munsterhjelm wrote: Your reference to a Cd variable to get population proportionality is interesting. I think part of the problem that you're trying to fix here comes from that clustering (such as by Lloyd's algorithm) optimizes for energy instead of mass. Or in other words, it finds districts so that the average distance for each voter to the closest center is minimized (energy) rather than to find districts that have equal proportions of people within their borders (analogous to equal mass, if each voter is a point of unit mass). Unless I'm mistaken, your generalization of Voronoi cells (with weights linked to the center nodes) is called power diagrams. Not long ago, I stumbled across a paper (draft?) about automatic redistricting involving power diagrams. It's here: http://www.stat.columbia.edu/%7Egelman/stuff_for_blog/fryer.pdf . The authors state that for a certain given compactness measure, the optimally compact district map will consist of power diagrams. They also give an algorithm for finding these optimal (or near- optimal, since the compactness measure is computationally intensive) power diagrams, but it's a bit too complex for my understanding. I'm looking at fryer.pdf and maybe the notation is a little thick for me but it looks like the pi(V_s) equation number 1 in section 3.2 is summing, for each district, the distance between every voter in the district and every other voter in that district. This is a little counterintuitive to me since for so long I've been summing the distance of voters to the center of their district. On the other hand, it does kinda make sense too. Anyway, what my pseudo-voronoi solver does is this: Initialization: • Assign district centers to random points (block coordinates). • Assign census blocks to nearest district center. Repeat until done: • Move district centers towards center of population they were actually assigned. • Adjust district weight, nudging weight up or down if there are too few or two many people assigned to it. • (optional) Nudge district centers away from underpopulated district centers and towards overpopulated district centers. • For each census block, assign census block to the district with the lowest ((distance to block from district center) * weightd) • Fixup district contiguity (because straight-line-nearest can jump across water and other gaps in odd ways) (That's now enshrined here: http://code.google.com/p/redistricter/wiki/NearestNeighborSolver ) Oh, and when it's done (an hour or two for most states), start it over again with a different random starting state because it may have gotten stuck in a local optimum. I think it's worth repeating that I'd prefer to write the law with a way of measuring what a good district mapping is, rather than writing the law with a specific procedure for creating the district mapping. Specific procedures get really complex, and I don't think in this case that a fair procedure ensures a fair outcome. Sounds reasonable. However, people don't really like randomness. There is considerable resistance to using randomness in elections and people might consider a system where the solution isn't well defined as somewhat random. You mean that the best solution mightn't be picked because nobody could find it? Also, there could be issues if a better solution is found after the deadline, especially, if it favoured one party over another. Perhaps the randomness complaint could be diminished by having a default map drawn according to some rules set in law. The redistricting law could refer to that rule, where the rule is very simple - perhaps splitline? Then a state commission or similar could draw the default map so that one is ensured that the results won't be too bad if the redistricting fails. As long as the rule is neutral, the state can't rig the default map, either. I guess my time in Computer Science land has left me pretty comfortable with the idea that there are lots of problems that are too hard to ever reliably get the best solution. I don't know if there's a short-short popularizing explanation of how finding a good solution is Hard while measuring the quality of a solution is pretty quick. If anybody asks and it's not the time, place, or audience for discussing NP Hard problems, I will wave my hands and say, Hey, look over there! Good results, on my one puny computer! With more it'd only get better! Brian Olson http://bolson.org/ Election-Methods mailing list - see http://electorama.com/em for list info
Re: [EM] A computationally feasible method
On Aug 31, 2008, at 8:25 AM, Raph Frank wrote: On Sat, Aug 30, 2008 at 5:46 PM, Juho [EMAIL PROTECTED] wrote: To gain even better trust that this set is the best one one could publish the best found set and then wait for a week and allow other interested parties to seek for even better sets. Maybe different parties or candidates try to find alternatives where they would do better. If nothing is found then the first found set is declared elected. Brian Olson suggests this approach for his anti-gerrymandering proposals. http://bolson.org/dist/USIRA.html and http://bolson.org/dist/ Ofc, he doesn't define geographic centers of the districts, which presumably means the centre of gravity of the district. I'm pretty sure I want the average point of the land area, but yes, there are different ways to count 'center' or 'middle' and there may be some debate between them. Maybe it would be better to define the centre of the district as the average position of all the people in the district. That's an obvious alternative, and it results in different shaped mappings when used, and it's not obvious which way is better, but I'm still leaning towards average distance per person to land-area-center rather than population-center. I think population center could be more likely to wind up with a million people right at center, and a few people flung off far away, but land-area-center is less likely for that to happen. One possible problem is that it would allow people with very powerful computers to gain an advantage. The Republicans and the Democrats would probably end up being favoured. However, the advantage is likely to be slight. Also, it could end up that there was a [EMAIL PROTECTED] like effort to find the 'true' best arrangement (or maybe both party's supporters doing their own version) [EMAIL PROTECTED] and [EMAIL PROTECTED] :) Given my anectodal experience with running solvers so far, it'll be pretty hard to get a good impartial score and also secretly bend the mapping towards some objective. This is substantially just because it's hard to get a good impartial score. I think the type of organization with the biggest chance of affecting the outcomes would be one with a big server farm: national labs with supercomputer clusters, Google, Yahoo, Amazon, NSA, etc. Election-Methods mailing list - see http://electorama.com/em for list info
[EM] language/framing quibble
I stopped reading the favoring racial minorities thread a long time ago, but every time I see the subject I cringe a little. Favoring sounds bad to me. I think it implies greater-than- proportional representation. racial minorities while often totally relevant, may be too narrow of a category when good PR could represent everyone of all social distinctions. If we're fixing something, we should strive towards accurate representation of previously underrepresented groups. No one can say anything against accurate representation, that's just good Democracy! previously underrepresented groups is inconveniently long, but is the best mix of connotation and denotation that I can think of right now. Just some things to think about so that once we've figured out what the best possible election method is, we'll be ready to switch into advocacy mode. Brian Olson http://bolson.org/ Election-Methods mailing list - see http://electorama.com/em for list info
Re: [Election-Methods] USING Condorcet
On Jun 30, 2008, at 9:51 AM, Dave Ketchum wrote: Condorcet provides for ranked approval for more than one candidate. This DOES NOT justify trying to get voters to rank more than they approve of. And, while I write above for voters to learn about other candidates, I do not see demanding that they try harder to learn about strays. It's worth rating everyone because if you wind up not getting any of the ones you 'approve of' you can still have some say in which of the rest of them you get. For the partially informed voter, that might be adjusted to say that it's worth voting on all the ones they have information on. On the 135 candidate ballot (CA-Gov 2003) I'd rank the 5-10 I had some opinion on, and not just the ones I like, but the ones I dislike in order of dislike because it could make a real difference to have Schwarzenegger over McClintock. Everyone here should get this immediately, and given that it can be summed up in a sentence, I think it's reasonable to try and add it to an educational pitch about Condorcet/Virtual-Round-Robin methods or at least have it ready to answer the Frequently Asked Questions. If trying to get voters to ... means forcing them to cast a full ranking or their ballot is invalid, that's bad. Educating them about their own self interest and how to cast the best vote possible, that's good. I imagine a voter education pamphlet with a line like: Voting for only one choice is bad. If that one doesn't win, you don't get any say over which of the others might get elected. Vote on as many choices as you feel informed enough to vote on. Brian Olson http://bolson.org/ Election-Methods mailing list - see http://electorama.com/em for list info
[Election-Methods] Bullet Voting in the wider media
In case anyone's interested in what the general public are hearing about voting strategy. http://www.boston.com/news/local/articles/2007/10/07/ ballot_query_to_bullet_or_not_to_bullet Election-Methods mailing list - see http://electorama.com/em for list info