okay, Sennet, I am posting this to the EM mailing list. ---------------------------- Original Message ---------------------------- Subject: Re: re Burlington From: "Sennet Williams" <sennetwilliams@yahoo.com> Date: Sat, May 18, 2019 12:50 pm To: "robert bristow-johnson" <rbj@audioimagination.com> -------------------------------------------------------------------------- > yes, I meant to put that in the email.  you are free to post it to the list. > As I probably typed before,  the problem with Condorcet is that it would be "practically" impossible to count by hand. No, Sennet, it isn't.  It's straight forward, but laborious.  If doing this by hand, you would need a team of 4 (2 callers and 2 counters) for each pair combination of candidates.  If you had 2 candidates, that's one pair (and it's just like FPTP)..  If you had 3 candidates, it's 3 pairs.  If you had 4 candidates, it's 6 pairs.  If you had 5 candidates, it's 10 pairs.  The counting could be done simultaneously if you had sufficient people or serially, in turn, if you don't have more enough for simultaneous counting.  all ballots would be handled by each counting team once.  and it is precinct summable so the burden can be distributed to many precinct locations.  unlike IRV, the counting need not be done at a single central location. but for a lot of candidates, like a dozen, IRV would be faster to do by hand, but still practical. >  In real elections, IRV, and Condorcet will have the same results:  The winning candidate will be the one who has the broadest preferred support. No, Sennet, that is decidedly false.  This is why i asked you if you really "understand what the difference is between IRV and Condorcet?"  When you make claims like that, it makes me wonder.  It's simply a demonstrably false assertion. The Burlington mayoral election in 2009 was a "real election".  Someone **really** got elected to office in that election. And IRV and Condorcet would have clearly gotten different results in that real election.  The IRV elected Bob Kiss.  And Condorcet would have elected Andy Montroll.  (And plurality of first-choice votes would have elected Kurt Wright.)  But it is only Condorcet that elects the candidate that is explicitly preferred by voters over every other candidate. > > On Friday, May 17, 2019, 11:40:35 AM PDT, robert bristow-johnson <rbj@audioimagination.com> wrote: > > > hi Sennet, > > can we post this to the list?  i didn't wanna do that without your consent.  it's just that maybe we can get someone else besides the two of us to pipe in on the conversation. > > > > ---------------------------- Original Message ---------------------------- > Subject: re Burlington > From: "Sennet Williams" <sennetwilliams@yahoo.com> > Date: Thu, May 16, 2019 8:54 pm > To: "rbj@audioimagination.com" <rbj@audioimagination.com> > -------------------------------------------------------------------------- > >> well, I have no idea what IRV system Burlington used. > > it's the same IRV as in every other governmental RCV election except we had 5 ranking levels and 5 candidates.  so no one was "disenfranchised".  you could have ranked the candidates in opposite order of their expost facto popularity, and you would still be able to weigh in on the IRV final round that actually selects the mayor. > > here is an analysis of what went wrong: https://rangevoting.org/Burlington.html > > here's another: http://bolson.org/~bolson/2009/20090303_burlington_vt_mayor.html > > essentially, we had 4 strong candidates going in.  3 were all plausible winners.  the GOP candidate had the Plurality, the Prog candidate wonthe IRV, and the Dem candidate was the pairwise champion.  the Dem would have beaten **any** other candidate in the IRV final round had he been able to advance to the final round.  that IRV eliminated him in the semi-final round was the execution of this inherent flaw of IRV. > > >> The problem we have had in SF, Berkeley and Oakland is that each voter can only select three candidates, and the number of exhausted ballots exceeded the winning margin in at least several elections. > > yes, that's a known problem with **any** RCV if the number of candidates on the ballotexceeds the number of ranking levels.  you need more ranking levels than 3 and you need stronger (or stricter) ballot access requirements so that fewer candidates get on the ballot and only those that are plausible winners.  i think 5 levels is enough, and the number of signatures on apetition needed to get on the ballot can be adjusted by law in response to the usual number of candidates that make it onto the ballot.  if there are consistently more names than ranking levels, the legislative body has the information and the authority necessary to increase the number ofrequired signatures to have candidate access to the ballot. > > >> Most clearly in the Kaplan/Quan/Perata mayor's contest (Oakland''s 1st IRV election)  There were also six "minor" candidates.  Kaplan was  almost surely the most preferred, but Quan gamed the system by mortgaging her house and spending a lot asking casual voters to"make me 2nd.  The winning margin over Kaplan was very narrow but the number of exhausted ballots was very large because most of the minor candidates were black while none of the big three were.   A lot of people blamed the IRV system for electing Quan, who was basicallyincompetent, but there has been no serious attempt to repeal IRV. > >   > > Ranked-Choice Voting will not stop bad politicians that are good salespersons from winning office.  But it is intended to stop spoiler candidates from preventing the candidate with the actual popular supportfrom winning. > > >> When CA gets statewide IRV, we would presumably Maine's system and all counties will be given new equipment so all candidates can be ranked. > > In Burlington we didn't need new equipment.  just new software.  the optical-scan machines were the same machines, but they had to beprogrammed slightly differently. > > Sennet, do you understand what the difference is between IRV and Condorcet?  What it is that we on the list bitch about regarding IRV. > > Our issue is not that we don't like RCV, we **want** Ranked-Choice Voting, we just want the rules reformed so thatthe pairwise champion is always elected.  IRV will do that *most* of the time, but it does not always do that.  and like the Electoral College, when IRV fails to elect who we all know should have been elected, it never brings legitimacy to the election.  failure to elect the pairwisechampion will only harm voting system reform. >   > > > -- > > r b-j                         rbj@audioimagination.com > > "Imagination is more important than knowledge." > >   >   > >   > >   >       -- r b-j                         rbj@audioimagination.com "Imagination is more important than knowledge."        

"But it is only Condorcet that elects the candidate that is explicitly preferred by voters over every other candidate." I wonder tho, whether that satisfies the requiremant for one candidate (of their number) to be prefered over a whole range of candidates? from Richard Lung. On 19/05/2019 01:30, robert bristow-johnson wrote: > > okay, Sennet, I am posting this to the EM mailing list. > > ---------------------------- Original Message ---------------------------- > Subject: Re: re Burlington > From: "Sennet Williams" <sennetwilliams@yahoo.com> > Date: Sat, May 18, 2019 12:50 pm > To: "robert bristow-johnson" <rbj@audioimagination.com> > -------------------------------------------------------------------------- > > > yes, I meant to put that in the email.  you are free to post it to > the list. > > As I probably typed before,  the problem with Condorcet is that it > would be "practically" impossible to count by hand. > > No, Sennet, it isn't.  It's straight forward, but laborious. If doing > this by hand, you would need a team of 4 (2 callers and 2 counters) > for each pair combination of candidates.  If you had 2 candidates, > that's one pair (and it's just like FPTP)..  If you had 3 candidates, > it's 3 pairs.  If you had 4 candidates, it's 6 pairs.  If you had 5 > candidates, it's 10 pairs.  The counting could be done simultaneously > if you had sufficient people or serially, in turn, if you don't have > more enough for simultaneous counting.  all ballots would be handled > by each counting team once.  and it is precinct summable so the burden > can be distributed to many precinct locations.  unlike IRV, the > counting need not be done at a single central location. > > but for a lot of candidates, like a dozen, IRV would be faster to do > by hand, but still practical. > > >  In real elections, IRV, and Condorcet will have the same results:  > The winning candidate will be the one who has the broadest preferred > support. > > No, Sennet, that is decidedly false.  This is why i asked you if you > really "understand what the difference is between IRV and Condorcet?"  > When you make claims like that, it makes me wonder.  It's simply a > demonstrably false assertion. > > The Burlington mayoral election in 2009 was a "real election". Someone > **really** got elected to office in that election. > > And IRV and Condorcet would have clearly gotten different results in > that real election.  The IRV elected Bob Kiss.  And Condorcet would > have elected Andy Montroll.  (And plurality of first-choice votes > would have elected Kurt Wright.)  But it is only Condorcet that elects > the candidate that is explicitly preferred by voters over every other > candidate. > > > > > > On Friday, May 17, 2019, 11:40:35 AM PDT, robert bristow-johnson > <rbj@audioimagination.com> wrote: > > > > > > hi Sennet, > > > > can we post this to the list?  i didn't wanna do that without your > consent.  it's just that maybe we can get someone else besides the two > of us to pipe in on the conversation. > > > > > > > > ---------------------------- Original Message > ---------------------------- > > Subject: re Burlington > > > From: "Sennet Williams" <sennetwilliams@yahoo.com> > > Date: Thu, May 16, 2019 8:54 pm > > To: "rbj@audioimagination.com" <rbj@audioimagination.com> > > > -------------------------------------------------------------------------- > > > >> well, I have no idea what IRV system Burlington used. > > > > it's the same IRV as in every other governmental RCV election except > we had 5 ranking levels and 5 candidates.  so no one was > "disenfranchised".  you could have ranked the candidates in opposite > order of their expost facto popularity, and you would still be able to > weigh in on the IRV final round that actually selects the mayor. > > > > here is an analysis of what went > wrong: https://rangevoting.org/Burlington.html > > > > here's > another: http://bolson.org/~bolson/2009/20090303_burlington_vt_mayor.html > > > > essentially, we had 4 strong candidates going in.  3 were all > plausible winners.  the GOP candidate had the Plurality, the Prog > candidate wonthe IRV, and the Dem candidate was the pairwise > champion.  the Dem would have beaten **any** other candidate in the > IRV final round had he been able to advance to the final round.  that > IRV eliminated him in the semi-final round was the execution of this > inherent flaw of IRV. > > > > > >> The problem we have had in SF, Berkeley and Oakland is that each > voter can only select three candidates, and the number of exhausted > ballots exceeded the winning margin in at least several elections. > > > > yes, that's a known problem with **any** RCV if the number of > candidates on the ballotexceeds the number of ranking levels.  you > need more ranking levels than 3 and you need stronger (or stricter) > ballot access requirements so that fewer candidates get on the ballot > and only those that are plausible winners.  i think 5 levels is > enough, and the number of signatures on apetition needed to get on the > ballot can be adjusted by law in response to the usual number of > candidates that make it onto the ballot.  if there are consistently > more names than ranking levels, the legislative body has the > information and the authority necessary to increase the number > ofrequired signatures to have candidate access to the ballot. > > > > > >> Most clearly in the Kaplan/Quan/Perata mayor's contest (Oakland''s > 1st IRV election)  There were also six "minor" candidates.  Kaplan > was  almost surely the most preferred, but Quan gamed the system by > mortgaging her house and spending a lot asking casual voters to"make > me 2nd.  The winning margin over Kaplan was very narrow but the number > of exhausted ballots was very large because most of the minor > candidates were black while none of the big three were.   A lot of > people blamed the IRV system for electing Quan, who was > basicallyincompetent, but there has been no serious attempt to repeal IRV. > > > > > > > > Ranked-Choice Voting will not stop bad politicians that are good > salespersons from winning office.  But it is intended to stop spoiler > candidates from preventing the candidate with the actual popular > supportfrom winning. > > > > > >> When CA gets statewide IRV, we would presumably Maine's system and > all counties will be given new equipment so all candidates can be ranked. > > > > In Burlington we didn't need new equipment.  just new software.  the > optical-scan machines were the same machines, but they had to > beprogrammed slightly differently. > > > > Sennet, do you understand what the difference is between IRV and > Condorcet?  What it is that we on the list bitch about regarding IRV. > > > > Our issue is not that we don't like RCV, we **want** Ranked-Choice > Voting, we just want the rules reformed so thatthe pairwise champion > is always elected.  IRV will do that *most* of the time, but it does > not always do that.  and like the Electoral College, when IRV fails to > elect who we all know should have been elected, it never brings > legitimacy to the election.  failure to elect the pairwisechampion > will only harm voting system reform. > > > > > > > > -- > > > > r b-j                         rbj@audioimagination.com > > > > "Imagination is more important than knowledge." > > > > > > > > > > > > > > > > > > > -- > > r b-j                         rbj@audioimagination.com > > "Imagination is more important than knowledge." > > > ---- > Election-Methods mailing list - see https://electorama.com/em for list info

---------------------------- Original Message ---------------------------- Subject: Re: [EM] re Burlington From: "Richard Lung" <voting@ukscientists.com> Date: Mon, May 20, 2019 11:52 am To: rbj@audioimagination.com election-methods@electorama.com Cc: "Sennet Williams" <sennetwilliams@yahoo.com> -------------------------------------------------------------------------- > "But it is only Condorcet that elects the candidate that is explicitly > preferred by voters over every other candidate." > > I wonder tho, whether that satisfies the requiremant for one candidate > (of their number) to be preferred over a whole range of candidates? > i am not sure exactly what you mean, here, Richard. if there are more than two candidates, while it might be possible for a single candidate to be preferred over the union of all the other candidates, that this single candidate has a *majority* of the vote, more first-choice votes than the sum of first-choice votes of all other candidates, while it may happen (and if it does happen, that single candidate must surely be elected given **any** deterministic voting system), but  i don't expect in a 3-way or 4-way race (or more) for that any candidate will satisfy that criterion. is that what you're asking about? but the Condorcet criterion simply proposes that if more voters mark their ballots preferring Candidate A to Candidate B than the number of voters marking their ballots to the contrary, then Candidate B is not elected. i like to compare it to a mathematical proof by contradiction: If a Condorcet Winner exists and your election method does not elect the CW, then who did you elect?  You elected someone when more voters explicitly marked their ballots that they preferred some other **specific** candidate.  How can that be an expression of the will of the electorate?  That's the problem with IRV or Borda or Bucklin (or some other RCV that someone cooks up).  Most of the time it may elect the CW.  Then great!  No one is complaining.  But when it doesn't, how possibly can the winner claim to have the democratic support of the electorate (even though we don't demand a majority of first-choice votes, since that demand is unrealistic in a 3-way or 4-way race) when the voters explicitly say "Give us this other candidate that we like better!" ?? it's sorta like this stupid Electoral College thing we have in the U.S.  if the Electoral College elects the same candidate as the Popular Vote does, we say that the E.C. does a pretty good job.  but when it *doesn't* elect the winner of the popular vote, it **never** brings legitimacy to the elected president and vice-president. L8r, r b-j > > On 19/05/2019 01:30, robert bristow-johnson wrote: >> >> okay, Sennet, I am posting this to the EM mailing list. >> >> ---------------------------- Original Message ---------------------------- >> Subject: Re: re Burlington >> From: "Sennet Williams" <sennetwilliams@yahoo.com> >> Date: Sat, May 18, 2019 12:50 pm >> To: "robert bristow-johnson" <rbj@audioimagination.com> >> -------------------------------------------------------------------------- >> >> > yes, I meant to put that in the email.  you are free to post it to >> the list. >> > As I probably typed before,  the problem with Condorcet is that it >> would be "practically" impossible to count by hand. >> >> No, Sennet, it isn't.  It's straight forward, but laborious. If doing >> this by hand, you would need a team of 4 (2 callers and 2 counters) >> for each pair combination of candidates.  If you had 2 candidates, >> that's one pair (and it's just like FPTP)..  If you had 3 candidates, >> it's 3 pairs.  If you had 4 candidates, it's 6 pairs.  If you had 5 >> candidates, it's 10 pairs.  The counting could be done simultaneously >> if you had sufficient people or serially, in turn, if you don't have >> more enough for simultaneous counting.  all ballots would be handled >> by each counting team once.  and it is precinct summable so the burden >> can be distributed to many precinct locations.  unlike IRV, the >> counting need not be done at a single central location. >> >> but for a lot of candidates, like a dozen, IRV would be faster to do >> by hand, but still practical. >> >> >  In real elections, IRV, and Condorcet will have the same results:  >> The winning candidate will be the one who has the broadest preferred >> support. >> >> No, Sennet, that is decidedly false.  This is why i asked you if you >> really "understand what the difference is between IRV and Condorcet?"  >> When you make claims like that, it makes me wonder.  It's simply a >> demonstrably false assertion. >> >> The Burlington mayoral election in 2009 was a "real election". Someone >> **really** got elected to office in that election. >> >> And IRV and Condorcet would have clearly gotten different results in >> that real election.  The IRV elected Bob Kiss.  And Condorcet would >> have elected Andy Montroll.  (And plurality of first-choice votes >> would have elected Kurt Wright.)  But it is only Condorcet that elects >> the candidate that is explicitly preferred by voters over every other >> candidate. >> >> >> > >> > On Friday, May 17, 2019, 11:40:35 AM PDT, robert bristow-johnson >> <rbj@audioimagination.com> wrote: >> > >> > >> > hi Sennet, >> > >> > can we post this to the list?  i didn't wanna do that without your >> consent.  it's just that maybe we can get someone else besides the two >> of us to pipe in on the conversation. >> > >> > >> > >> > ---------------------------- Original Message >> ---------------------------- >> > Subject: re Burlington >> > >> From: "Sennet Williams" <sennetwilliams@yahoo.com> >> > Date: Thu, May 16, 2019 8:54 pm >> > To: "rbj@audioimagination.com" <rbj@audioimagination.com> >> > >> -------------------------------------------------------------------------- >> > >> >> well, I have no idea what IRV system Burlington used. >> > >> > it's the same IRV as in every other governmental RCV election except >> we had 5 ranking levels and 5 candidates.  so no one was >> "disenfranchised".  you could have ranked the candidates in opposite >> order of their expost facto popularity, and you would still be able to >> weigh in on the IRV final round that actually selects the mayor. >> > >> > here is an analysis of what went >> wrong: https://rangevoting.org/Burlington.html >> > >> > here's >> another: http://bolson.org/~bolson/2009/20090303_burlington_vt_mayor.html >> > >> > essentially, we had 4 strong candidates going in.  3 were all >> plausible winners.  the GOP candidate had the Plurality, the Prog >> candidate wonthe IRV, and the Dem candidate was the pairwise >> champion.  the Dem would have beaten **any** other candidate in the >> IRV final round had he been able to advance to the final round.  that >> IRV eliminated him in the semi-final round was the execution of this >> inherent flaw of IRV. >> > >> > >> >> The problem we have had in SF, Berkeley and Oakland is that each >> voter can only select three candidates, and the number of exhausted >> ballots exceeded the winning margin in at least several elections. >> > >> > yes, that's a known problem with **any** RCV if the number of >> candidates on the ballotexceeds the number of ranking levels.  you >> need more ranking levels than 3 and you need stronger (or stricter) >> ballot access requirements so that fewer candidates get on the ballot >> and only those that are plausible winners.  i think 5 levels is >> enough, and the number of signatures on apetition needed to get on the >> ballot can be adjusted by law in response to the usual number of >> candidates that make it onto the ballot.  if there are consistently >> more names than ranking levels, the legislative body has the >> information and the authority necessary to increase the number >> ofrequired signatures to have candidate access to the ballot. >> > >> > >> >> Most clearly in the Kaplan/Quan/Perata mayor's contest (Oakland''s >> 1st IRV election)  There were also six "minor" candidates.  Kaplan >> was  almost surely the most preferred, but Quan gamed the system by >> mortgaging her house and spending a lot asking casual voters to"make >> me 2nd.  The winning margin over Kaplan was very narrow but the number >> of exhausted ballots was very large because most of the minor >> candidates were black while none of the big three were.   A lot of >> people blamed the IRV system for electing Quan, who was >> basicallyincompetent, but there has been no serious attempt to repeal IRV. >> > >> > >> > >> > Ranked-Choice Voting will not stop bad politicians that are good >> salespersons from winning office.  But it is intended to stop spoiler >> candidates from preventing the candidate with the actual popular >> supportfrom winning. >> > >> > >> >> When CA gets statewide IRV, we would presumably Maine's system and >> all counties will be given new equipment so all candidates can be ranked. >> > >> > In Burlington we didn't need new equipment.  just new software.  the >> optical-scan machines were the same machines, but they had to >> beprogrammed slightly differently. >> > >> > Sennet, do you understand what the difference is between IRV and >> Condorcet?  What it is that we on the list bitch about regarding IRV. >> > >> > Our issue is not that we don't like RCV, we **want** Ranked-Choice >> Voting, we just want the rules reformed so thatthe pairwise champion >> is always elected.  IRV will do that *most* of the time, but it does >> not always do that.  and like the Electoral College, when IRV fails to >> elect who we all know should have been elected, it never brings >> legitimacy to the election.  failure to elect the pairwisechampion >> will only harm voting system reform. >> > >> > >> > >> > -- >> > >> > r b-j                         rbj@audioimagination.com >> > >> > "Imagination is more important than knowledge." >> >      

Hello Robert and all, I wasn't sure what I meant but I think now what I'm getting at is that a Condorcet winner may not necessarily have reached an adequate level of support, in terms of a quota requirement. (Another message seems to touch on a similar point.) Proving adequate support would imply over-coming something like a NOTA option, tho I don't think that should be an explicit option on the ballot paper, because it amounts to privileging one preference permutation out of many (which is also what is inequitable about party list systems). However, I believe all preference abstentions should be counted (using all the preference information, not just some of it). I assume that objections to IRV are based on its "premature exclusion" of candidates, which undermines its credibility as a quota count. This is less true in multi-member constituencies, where there are equitably transferable votes to elect more than one representative. However, the objection can be completely over-come, if premature exclusion is replaced by a rational count of votes in order of the least prefered candidates. Previously mentioned, I have worked out how to do this in a coherent new electoral system. from Richard Lung. On 20/05/2019 21:12, robert bristow-johnson wrote: > > > > ---------------------------- Original Message ---------------------------- > Subject: Re: [EM] re Burlington > From: "Richard Lung" <voting@ukscientists.com> > Date: Mon, May 20, 2019 11:52 am > To: rbj@audioimagination.com > election-methods@electorama.com > Cc: "Sennet Williams" <sennetwilliams@yahoo.com> > -------------------------------------------------------------------------- > > > "But it is only Condorcet that elects the candidate that is explicitly > > preferred by voters over every other candidate." > > > > I wonder tho, whether that satisfies the requiremant for one candidate > > (of their number) to be preferred over a whole range of candidates? > > > > i am not sure exactly what you mean, here, Richard. > > if there are more than two candidates, while it might be possible for > a single candidate to be preferred over the union of all the other > candidates, that this single candidate has a *majority* of the vote, > more first-choice votes than the sum of first-choice votes of all > other candidates, while it may happen (and if it does happen, that > single candidate must surely be elected given **any** deterministic > voting system), but  i don't expect in a 3-way or 4-way race (or more) > for that any candidate will satisfy that criterion. > > is that what you're asking about? > > but the Condorcet criterion simply proposes that if more voters mark > their ballots preferring Candidate A to Candidate B than the number of > voters marking their ballots to the contrary, then Candidate B is not > elected. > > i like to compare it to a mathematical proof by contradiction: If a > Condorcet Winner exists and your election method does not elect the > CW, then who did you elect?  You elected someone when more voters > explicitly marked their ballots that they preferred some other > **specific** candidate.  How can that be an expression of the will of > the electorate?  That's the problem with IRV or Borda or Bucklin (or > some other RCV that someone cooks up).  Most of the time it may elect > the CW.  Then great! No one is complaining.  But when it doesn't, how > possibly can the winner claim to have the democratic support of the > electorate (even though we don't demand a majority of first-choice > votes, since that demand is unrealistic in a 3-way or 4-way race) when > the voters explicitly say "Give us this other candidate that we like > better!" ?? > > it's sorta like this stupid Electoral College thing we have in the > U.S.  if the Electoral College elects the same candidate as the > Popular Vote does, we say that the E.C. does a pretty good job.  but > when it *doesn't* elect the winner of the popular vote, it **never** > brings legitimacy to the elected president and vice-president. > > L8r, > > r b-j > > > > > > On 19/05/2019 01:30, robert bristow-johnson wrote: > >> > >> okay, Sennet, I am posting this to the EM mailing list. > >> > >> ---------------------------- Original Message > ---------------------------- > >> Subject: Re: re Burlington > >> > From: "Sennet Williams" <sennetwilliams@yahoo.com> > >> Date: Sat, May 18, 2019 12:50 pm > >> To: "robert bristow-johnson" <rbj@audioimagination.com> > >> > -------------------------------------------------------------------------- > >> > >> > yes, I meant to put that in the email.  you are free to post it to > >> the list. > >> > As I probably typed before,  the problem with Condorcet is that it > >> would be "practically" impossible to count by hand. > >> > >> No, Sennet, it isn't.  It's straight forward, but laborious. If doing > >> this by hand, you would need a team of 4 (2 callers and 2 counters) > >> for each pair combination of candidates.  If you had 2 candidates, > >> that's one pair (and it's just like FPTP)..  If you had 3 candidates, > >> it's 3 pairs.  If you had 4 candidates, it's 6 pairs. If you had 5 > >> candidates, it's 10 pairs.  The counting could be done simultaneously > >> if you had sufficient people or serially, in turn, if you don't have > >> more enough for simultaneous counting.  all ballots would be handled > >> by each counting team once.  and it is precinct summable so the burden > >> can be distributed to many precinct locations.  unlike IRV, the > >> counting need not be done at a single central location. > >> > >> but for a lot of candidates, like a dozen, IRV would be faster to do > >> by hand, but still practical. > >> > >> >  In real elections, IRV, and Condorcet will have the same results: > >> The winning candidate will be the one who has the broadest preferred > >> support. > >> > >> No, Sennet, that is decidedly false.  This is why i asked you if you > >> really "understand what the difference is between IRV and Condorcet?" > >> When you make claims like that, it makes me wonder. It's simply a > >> demonstrably false assertion. > >> > >> The Burlington mayoral election in 2009 was a "real election". Someone > >> **really** got elected to office in that election. > >> > >> And IRV and Condorcet would have clearly gotten different results in > >> that real election.  The IRV elected Bob Kiss.  And Condorcet would > >> have elected Andy Montroll.  (And plurality of first-choice votes > >> would have elected Kurt Wright.)  But it is only Condorcet that elects > >> the candidate that is explicitly preferred by voters over every other > >> candidate. > >> > >> > >> > > >> > On Friday, May 17, 2019, 11:40:35 AM PDT, robert bristow-johnson > >> <rbj@audioimagination.com> wrote: > >> > > >> > > >> > hi Sennet, > >> > > >> > can we post this to the list?  i didn't wanna do that without your > >> consent.  it's just that maybe we can get someone else besides the two > >> of us to pipe in on the conversation. > >> > > >> > > >> > > >> > ---------------------------- Original Message > >> ---------------------------- > >> > Subject: re Burlington > >> > > >> > From: "Sennet Williams" <sennetwilliams@yahoo.com> > >> > Date: Thu, May 16, 2019 8:54 pm > >> > To: "rbj@audioimagination.com" <rbj@audioimagination.com> > >> > > >> > -------------------------------------------------------------------------- > >> > > >> >> well, I have no idea what IRV system Burlington used. > >> > > >> > it's the same IRV as in every other governmental RCV election except > >> we had 5 ranking levels and 5 candidates.  so no one was > >> "disenfranchised".  you could have ranked the candidates in opposite > >> order of their expost facto popularity, and you would still be able to > >> weigh in on the IRV final round that actually selects the mayor. > >> > > >> > here is an analysis of what went > >> wrong: https://rangevoting.org/Burlington.html > >> > > >> > here's > >> > another: http://bolson.org/~bolson/2009/20090303_burlington_vt_mayor.html > >> > > >> > essentially, we had 4 strong candidates going in. 3 were all > >> plausible winners.  the GOP candidate had the Plurality, the Prog > >> candidate wonthe IRV, and the Dem candidate was the pairwise > >> champion.  the Dem would have beaten **any** other candidate in the > >> IRV final round had he been able to advance to the final round.  that > >> IRV eliminated him in the semi-final round was the execution of this > >> inherent flaw of IRV. > >> > > >> > > >> >> The problem we have had in SF, Berkeley and Oakland is that each > >> voter can only select three candidates, and the number of exhausted > >> ballots exceeded the winning margin in at least several elections. > >> > > >> > yes, that's a known problem with **any** RCV if the number of > >> candidates on the ballotexceeds the number of ranking levels.  you > >> need more ranking levels than 3 and you need stronger (or stricter) > >> ballot access requirements so that fewer candidates get on the ballot > >> and only those that are plausible winners.  i think 5 levels is > >> enough, and the number of signatures on apetition needed to get on the > >> ballot can be adjusted by law in response to the usual number of > >> candidates that make it onto the ballot.  if there are consistently > >> more names than ranking levels, the legislative body has the > >> information and the authority necessary to increase the number > >> ofrequired signatures to have candidate access to the ballot. > >> > > >> > > >> >> Most clearly in the Kaplan/Quan/Perata mayor's contest (Oakland''s > >> 1st IRV election)  There were also six "minor" candidates.  Kaplan > >> was  almost surely the most preferred, but Quan gamed the system by > >> mortgaging her house and spending a lot asking casual voters to"make > >> me 2nd.  The winning margin over Kaplan was very narrow but the number > >> of exhausted ballots was very large because most of the minor > >> candidates were black while none of the big three were.   A lot of > >> people blamed the IRV system for electing Quan, who was > >> basicallyincompetent, but there has been no serious attempt to > repeal IRV. > >> > > >> > > >> > > >> > Ranked-Choice Voting will not stop bad politicians that are good > >> salespersons from winning office.  But it is intended to stop spoiler > >> candidates from preventing the candidate with the actual popular > >> supportfrom winning. > >> > > >> > > >> >> When CA gets statewide IRV, we would presumably Maine's system and > >> all counties will be given new equipment so all candidates can be > ranked. > >> > > >> > In Burlington we didn't need new equipment.  just new software.  the > >> optical-scan machines were the same machines, but they had to > >> beprogrammed slightly differently. > >> > > >> > Sennet, do you understand what the difference is between IRV and > >> Condorcet?  What it is that we on the list bitch about regarding IRV. > >> > > >> > Our issue is not that we don't like RCV, we **want** Ranked-Choice > >> Voting, we just want the rules reformed so thatthe pairwise champion > >> is always elected.  IRV will do that *most* of the time, but it does > >> not always do that.  and like the Electoral College, when IRV fails to > >> elect who we all know should have been elected, it never brings > >> legitimacy to the election.  failure to elect the pairwisechampion > >> will only harm voting system reform. > >> > > >> > > >> > > >> > -- > >> > > >> > r b-j  rbj@audioimagination.com > >> > > >> > "Imagination is more important than knowledge." > >> > > > > ---- > Election-Methods mailing list - see https://electorama.com/em for list info

Re: > The problem we have had in SF, Berkeley and Oakland is that each > >> voter can only select three candidates, and the number of exhausted > >> ballots exceeded the winning margin in at least several elections. > >> > > >> > yes, that's a known problem with **any** RCV if the number of > >> candidates on the ballotexceeds the number of ranking levels.?? you > >> need more ranking levels than 3 and you need stronger (or stricter) > >> ballot access requirements so that fewer candidates get on the ballot > >> and only those that are plausible winners.?? i think 5 levels is > >> enough, and the number of signatures on apetition needed to get on the > >> ballot can be adjusted by law in response to the usual number of > >> candidates that make it onto the ballot.?? if there are consistently > >> more names than ranking levels, the legislative body has the > >> information and the authority necessary to increase the number > >> ofrequired signatures to have candidate access to the ballot. > >> > > >> > > >> >> Most clearly in the Kaplan/Quan/Perata mayor's contest (Oakland''s > >> 1st IRV election)?? There were also six "minor" candidates.?? Kaplan > >> was?? almost surely the most preferred, but Quan gamed the system by > >> mortgaging her house and spending a lot asking casual voters to"make > >> me 2nd.?? The winning margin over Kaplan was very narrow but the number > >> of exhausted ballots was very large because most of the minor > >> candidates were black while none of the big three were.???? A lot of > >> people blamed the IRV system for electing Quan, who was > >> basicallyincompetent, but there has been no serious attempt to > repeal IRV. It still boggles my mind that anyone would think it appropriate to not allow voters to strictly rank as many candidates as they wish. Here in Australia we just had a Federal election. Voting is compulsory and for the single-member districts of the House of Representatives voters must strictly rank all the candidates by writing numbers in boxes on paper ballots. If they fail to do that their vote is classified as "informal" and?? has no effect on the result. The average number of candidates per single-member district?? was 7.?? I'm having trouble finding out the maximum number but at least one had 13.?? The total "informal vote" was 5.49%. (BTW, I am also strongly of the view that voters should be allowed to rank as *few* candidates as they wish.) There are no petitions to gain ballot access. Instead you just put up a cash deposit of?? A$2000 and it is returned to you if you are elected or get at least 4% of?? the first-preference vote. Chris Benham On 21/05/2019 5:42 am, robert bristow-johnson wrote: > > > > ---------------------------- Original Message ---------------------------- > Subject: Re: [EM] re Burlington > From: "Richard Lung" <voting@ukscientists.com> > Date: Mon, May 20, 2019 11:52 am > To: rbj@audioimagination.com > election-methods@electorama.com > Cc: "Sennet Williams" <sennetwilliams@yahoo.com> > -------------------------------------------------------------------------- > > > "But it is only Condorcet that elects the candidate that is explicitly > > preferred by voters over every other candidate." > > > > I wonder tho, whether that satisfies the requiremant for one candidate > > (of their number) to be preferred over a whole range of candidates? > > > > i am not sure exactly what you mean, here, Richard. > > if there are more than two candidates, while it might be possible for > a single candidate to be preferred over the union of all the other > candidates, that this single candidate has a *majority* of the vote, > more first-choice votes than the sum of first-choice votes of all > other candidates, while it may happen (and if it does happen, that > single candidate must surely be elected given **any** deterministic > voting system), but?? i don't expect in a 3-way or 4-way race (or more) > for that any candidate will satisfy that criterion. > > is that what you're asking about? > > but the Condorcet criterion simply proposes that if more voters mark > their ballots preferring Candidate A to Candidate B than the number of > voters marking their ballots to the contrary, then Candidate B is not > elected. > > i like to compare it to a mathematical proof by contradiction: If a > Condorcet Winner exists and your election method does not elect the > CW, then who did you elect??? You elected someone when more voters > explicitly marked their ballots that they preferred some other > **specific** candidate.?? How can that be an expression of the will of > the electorate??? That's the problem with IRV or Borda or Bucklin (or > some other RCV that someone cooks up).?? Most of the time it may elect > the CW.?? Then great! No one is complaining.?? But when it doesn't, how > possibly can the winner claim to have the democratic support of the > electorate (even though we don't demand a majority of first-choice > votes, since that demand is unrealistic in a 3-way or 4-way race) when > the voters explicitly say "Give us this other candidate that we like > better!" ?? > > it's sorta like this stupid Electoral College thing we have in the > U.S.?? if the Electoral College elects the same candidate as the > Popular Vote does, we say that the E.C. does a pretty good job.?? but > when it *doesn't* elect the winner of the popular vote, it **never** > brings legitimacy to the elected president and vice-president. > > L8r, > > r b-j > > > > > > On 19/05/2019 01:30, robert bristow-johnson wrote: > >> > >> okay, Sennet, I am posting this to the EM mailing list. > >> > >> ---------------------------- Original Message > ---------------------------- > >> Subject: Re: re Burlington > >> > From: "Sennet Williams" <sennetwilliams@yahoo.com> > >> Date: Sat, May 18, 2019 12:50 pm > >> To: "robert bristow-johnson" <rbj@audioimagination.com> > >> > -------------------------------------------------------------------------- > >> > >> > yes, I meant to put that in the email.?? you are free to post it to > >> the list. > >> > As I probably typed before,?? the problem with Condorcet is that it > >> would be "practically" impossible to count by hand. > >> > >> No, Sennet, it isn't.?? It's straight forward, but laborious. If doing > >> this by hand, you would need a team of 4 (2 callers and 2 counters) > >> for each pair combination of candidates.?? If you had 2 candidates, > >> that's one pair (and it's just like FPTP)..?? If you had 3 candidates, > >> it's 3 pairs.?? If you had 4 candidates, it's 6 pairs. If you had 5 > >> candidates, it's 10 pairs.?? The counting could be done simultaneously > >> if you had sufficient people or serially, in turn, if you don't have > >> more enough for simultaneous counting.?? all ballots would be handled > >> by each counting team once.?? and it is precinct summable so the burden > >> can be distributed to many precinct locations.?? unlike IRV, the > >> counting need not be done at a single central location. > >> > >> but for a lot of candidates, like a dozen, IRV would be faster to do > >> by hand, but still practical. > >> > >> >?? In real elections, IRV, and Condorcet will have the same results: > >> The winning candidate will be the one who has the broadest preferred > >> support. > >> > >> No, Sennet, that is decidedly false.?? This is why i asked you if you > >> really "understand what the difference is between IRV and Condorcet?" > >> When you make claims like that, it makes me wonder. It's simply a > >> demonstrably false assertion. > >> > >> The Burlington mayoral election in 2009 was a "real election". Someone > >> **really** got elected to office in that election. > >> > >> And IRV and Condorcet would have clearly gotten different results in > >> that real election.?? The IRV elected Bob Kiss.?? And Condorcet would > >> have elected Andy Montroll.?? (And plurality of first-choice votes > >> would have elected Kurt Wright.)?? But it is only Condorcet that elects > >> the candidate that is explicitly preferred by voters over every other > >> candidate. > >> > >> > >> > > >> > On Friday, May 17, 2019, 11:40:35 AM PDT, robert bristow-johnson > >> <rbj@audioimagination.com> wrote: > >> > > >> > > >> > hi Sennet, > >> > > >> > can we post this to the list??? i didn't wanna do that without your > >> consent.?? it's just that maybe we can get someone else besides the two > >> of us to pipe in on the conversation. > >> > > >> > > >> > > >> > ---------------------------- Original Message > >> ---------------------------- > >> > Subject: re Burlington > >> > > >> > From: "Sennet Williams" <sennetwilliams@yahoo.com> > >> > Date: Thu, May 16, 2019 8:54 pm > >> > To: "rbj@audioimagination.com" <rbj@audioimagination.com> > >> > > >> > -------------------------------------------------------------------------- > >> > > >> >> well, I have no idea what IRV system Burlington used. > >> > > >> > it's the same IRV as in every other governmental RCV election except > >> we had 5 ranking levels and 5 candidates.?? so no one was > >> "disenfranchised".?? you could have ranked the candidates in opposite > >> order of their expost facto popularity, and you would still be able to > >> weigh in on the IRV final round that actually selects the mayor. > >> > > >> > here is an analysis of what went > >> wrong:??https://rangevoting.org/Burlington.html > >> > > >> > here's > >> > another:??http://bolson.org/~bolson/2009/20090303_burlington_vt_mayor.html > >> > > >> > essentially, we had 4 strong candidates going in. 3 were all > >> plausible winners.?? the GOP candidate had the Plurality, the Prog > >> candidate wonthe IRV, and the Dem candidate was the pairwise > >> champion.?? the Dem would have beaten **any** other candidate in the > >> IRV final round had he been able to advance to the final round.?? that > >> IRV eliminated him in the semi-final round was the execution of this > >> inherent flaw of IRV. > >> > > >> > > >> >> The problem we have had in SF, Berkeley and Oakland is that each > >> voter can only select three candidates, and the number of exhausted > >> ballots exceeded the winning margin in at least several elections. > >> > > >> > yes, that's a known problem with **any** RCV if the number of > >> candidates on the ballotexceeds the number of ranking levels.?? you > >> need more ranking levels than 3 and you need stronger (or stricter) > >> ballot access requirements so that fewer candidates get on the ballot > >> and only those that are plausible winners.?? i think 5 levels is > >> enough, and the number of signatures on apetition needed to get on the > >> ballot can be adjusted by law in response to the usual number of > >> candidates that make it onto the ballot.?? if there are consistently > >> more names than ranking levels, the legislative body has the > >> information and the authority necessary to increase the number > >> ofrequired signatures to have candidate access to the ballot. > >> > > >> > > >> >> Most clearly in the Kaplan/Quan/Perata mayor's contest (Oakland''s > >> 1st IRV election)?? There were also six "minor" candidates.?? Kaplan > >> was?? almost surely the most preferred, but Quan gamed the system by > >> mortgaging her house and spending a lot asking casual voters to"make > >> me 2nd.?? The winning margin over Kaplan was very narrow but the number > >> of exhausted ballots was very large because most of the minor > >> candidates were black while none of the big three were.???? A lot of > >> people blamed the IRV system for electing Quan, who was > >> basicallyincompetent, but there has been no serious attempt to > repeal IRV. > >> > > >> > > >> > > >> > Ranked-Choice Voting will not stop bad politicians that are good > >> salespersons from winning office.?? But it is intended to stop spoiler > >> candidates from preventing the candidate with the actual popular > >> supportfrom winning. > >> > > >> > > >> >> When CA gets statewide IRV, we would presumably Maine's system and > >> all counties will be given new equipment so all candidates can be > ranked. > >> > > >> > In Burlington we didn't need new equipment.?? just new software.?? the > >> optical-scan machines were the same machines, but they had to > >> beprogrammed slightly differently. > >> > > >> > Sennet, do you understand what the difference is between IRV and > >> Condorcet??? What it is that we on the list bitch about regarding IRV. > >> > > >> > Our issue is not that we don't like RCV, we **want** Ranked-Choice > >> Voting, we just want the rules reformed so thatthe pairwise champion > >> is always elected.?? IRV will do that *most* of the time, but it does > >> not always do that.?? and like the Electoral College, when IRV fails to > >> elect who we all know should have been elected, it never brings > >> legitimacy to the election.?? failure to elect the pairwisechampion > >> will only harm voting system reform. > >> > > >> > > >> > > >> > -- > >> > > >> > r b-j ??rbj@audioimagination.com > >> > > >> > "Imagination is more important than knowledge." > >> > > > > ---- > Election-Methods mailing list - see https://electorama.com/em for list info --- This email has been checked for viruses by AVG. https://www.avg.com

On 5/22/2019 9:34 AM, Chris Benham wrote: > It still boggles my mind that anyone would think it appropriate to not > allow voters to strictly rank > as many candidates as they wish. An explanation is in order for list members outside the United States. Essentially all ballots here are processed using optical scan and/or touchscreen voting machines. Limitations on the number of candidates the voter can rank are the direct result of the antiquated voting machines still in use. Accepting these limitations has been a necessary price of getting any ranked ballot method adopted anywhere in the U.S. No one thinks they are a good idea, except some opponents of reform who argue for a limited number of rankings as a kind of rear-guard resistance to the reform itself. The voting machine situation is improving, but is still not perfect. Current systems from the three main vendors can accommodate about 20, 10 and 6 rankings respectively. Six will become 10 as soon as that vendor's software can print and scan landscape ballots. Practical reformers take the understandable position that the perfect should not be allowed to be the enemy of the good. --Bob Richard On 5/22/2019 9:34 AM, Chris Benham wrote: > > Re: > >> The problem we have had in SF, Berkeley and Oakland is that each >> >> voter can only select three candidates, and the number of exhausted >> >> ballots exceeded the winning margin in at least several elections. >> >> > >> >> > yes, that's a known problem with **any** RCV if the number of >> >> candidates on the ballotexceeds the number of ranking levels.?? you >> >> need more ranking levels than 3 and you need stronger (or stricter) >> >> ballot access requirements so that fewer candidates get on the ballot >> >> and only those that are plausible winners.?? i think 5 levels is >> >> enough, and the number of signatures on apetition needed to get on the >> >> ballot can be adjusted by law in response to the usual number of >> >> candidates that make it onto the ballot.?? if there are consistently >> >> more names than ranking levels, the legislative body has the >> >> information and the authority necessary to increase the number >> >> ofrequired signatures to have candidate access to the ballot. >> >> > >> >> > >> >> >> Most clearly in the Kaplan/Quan/Perata mayor's contest (Oakland''s >> >> 1st IRV election)?? There were also six "minor" candidates.?? Kaplan >> >> was?? almost surely the most preferred, but Quan gamed the system by >> >> mortgaging her house and spending a lot asking casual voters to"make >> >> me 2nd.?? The winning margin over Kaplan was very narrow but the >> number >> >> of exhausted ballots was very large because most of the minor >> >> candidates were black while none of the big three were.???? A lot of >> >> people blamed the IRV system for electing Quan, who was >> >> basicallyincompetent, but there has been no serious attempt to >> repeal IRV. > > It still boggles my mind that anyone would think it appropriate to not > allow voters to strictly rank > as many candidates as they wish. > > Here in Australia we just had a Federal election. Voting is compulsory > and for the single-member districts > of the House of Representatives voters must strictly rank all the > candidates by writing numbers in boxes > on paper ballots. If they fail to do that their vote is classified as > "informal" and?? has no effect on the result. > > The average number of candidates per single-member district?? was 7.?? > I'm having trouble finding out the maximum > number but at least one had 13.?? The total "informal vote" was 5.49%. > > (BTW, I am also strongly of the view that voters should be allowed to > rank as *few* candidates as they wish.) > > There are no petitions to gain ballot access. Instead you just put up > a cash deposit of?? A$2000 and it is returned to > you if you are elected or get at least 4% of?? the first-preference vote. > > Chris Benham > > On 21/05/2019 5:42 am, robert bristow-johnson wrote: >> >> >> >> ---------------------------- Original Message >> ---------------------------- >> Subject: Re: [EM] re Burlington >> From: "Richard Lung" <voting@ukscientists.com> >> Date: Mon, May 20, 2019 11:52 am >> To: rbj@audioimagination.com >> election-methods@electorama.com >> Cc: "Sennet Williams" <sennetwilliams@yahoo.com> >> -------------------------------------------------------------------------- >> >> > "But it is only Condorcet that elects the candidate that is explicitly >> > preferred by voters over every other candidate." >> > >> > I wonder tho, whether that satisfies the requiremant for one candidate >> > (of their number) to be preferred over a whole range of candidates? >> > >> >> i am not sure exactly what you mean, here, Richard. >> >> if there are more than two candidates, while it might be possible for >> a single candidate to be preferred over the union of all the other >> candidates, that this single candidate has a *majority* of the vote, >> more first-choice votes than the sum of first-choice votes of all >> other candidates, while it may happen (and if it does happen, that >> single candidate must surely be elected given **any** deterministic >> voting system), but?? i don't expect in a 3-way or 4-way race (or >> more) for that any candidate will satisfy that criterion. >> >> is that what you're asking about? >> >> but the Condorcet criterion simply proposes that if more voters mark >> their ballots preferring Candidate A to Candidate B than the number >> of voters marking their ballots to the contrary, then Candidate B is >> not elected. >> >> i like to compare it to a mathematical proof by contradiction: If a >> Condorcet Winner exists and your election method does not elect the >> CW, then who did you elect??? You elected someone when more voters >> explicitly marked their ballots that they preferred some other >> **specific** candidate.?? How can that be an expression of the will >> of the electorate??? That's the problem with IRV or Borda or Bucklin >> (or some other RCV that someone cooks up).?? Most of the time it may >> elect the CW.?? Then great!?? No one is complaining.?? But when it >> doesn't, how possibly can the winner claim to have the democratic >> support of the electorate (even though we don't demand a majority of >> first-choice votes, since that demand is unrealistic in a 3-way or >> 4-way race) when the voters explicitly say "Give us this other >> candidate that we like better!" ?? >> >> it's sorta like this stupid Electoral College thing we have in the >> U.S.?? if the Electoral College elects the same candidate as the >> Popular Vote does, we say that the E.C. does a pretty good job.?? but >> when it *doesn't* elect the winner of the popular vote, it **never** >> brings legitimacy to the elected president and vice-president. >> >> L8r, >> >> r b-j >> >> >> > >> > On 19/05/2019 01:30, robert bristow-johnson wrote: >> >> >> >> okay, Sennet, I am posting this to the EM mailing list. >> >> >> >> ---------------------------- Original Message >> ---------------------------- >> >> Subject: Re: re Burlington >> >> >> From: "Sennet Williams" <sennetwilliams@yahoo.com> >> >> Date: Sat, May 18, 2019 12:50 pm >> >> To: "robert bristow-johnson" <rbj@audioimagination.com> >> >> >> -------------------------------------------------------------------------- >> >> >> >> > yes, I meant to put that in the email.?? you are free to post it to >> >> the list. >> >> > As I probably typed before,?? the problem with Condorcet is that it >> >> would be "practically" impossible to count by hand. >> >> >> >> No, Sennet, it isn't.?? It's straight forward, but laborious. If doing >> >> this by hand, you would need a team of 4 (2 callers and 2 counters) >> >> for each pair combination of candidates.?? If you had 2 candidates, >> >> that's one pair (and it's just like FPTP)..?? If you had 3 candidates, >> >> it's 3 pairs.?? If you had 4 candidates, it's 6 pairs.?? If you had 5 >> >> candidates, it's 10 pairs.?? The counting could be done simultaneously >> >> if you had sufficient people or serially, in turn, if you don't have >> >> more enough for simultaneous counting.?? all ballots would be handled >> >> by each counting team once.?? and it is precinct summable so the >> burden >> >> can be distributed to many precinct locations.?? unlike IRV, the >> >> counting need not be done at a single central location. >> >> >> >> but for a lot of candidates, like a dozen, IRV would be faster to do >> >> by hand, but still practical. >> >> >> >> >?? In real elections, IRV, and Condorcet will have the same >> results:?? >> >> The winning candidate will be the one who has the broadest preferred >> >> support. >> >> >> >> No, Sennet, that is decidedly false.?? This is why i asked you if you >> >> really "understand what the difference is between IRV and >> Condorcet?"?? >> >> When you make claims like that, it makes me wonder.?? It's simply a >> >> demonstrably false assertion. >> >> >> >> The Burlington mayoral election in 2009 was a "real election". Someone >> >> **really** got elected to office in that election. >> >> >> >> And IRV and Condorcet would have clearly gotten different results in >> >> that real election.?? The IRV elected Bob Kiss.?? And Condorcet would >> >> have elected Andy Montroll.?? (And plurality of first-choice votes >> >> would have elected Kurt Wright.)?? But it is only Condorcet that >> elects >> >> the candidate that is explicitly preferred by voters over every other >> >> candidate. >> >> >> >> >> >> > >> >> > On Friday, May 17, 2019, 11:40:35 AM PDT, robert bristow-johnson >> >> <rbj@audioimagination.com> wrote: >> >> > >> >> > >> >> > hi Sennet, >> >> > >> >> > can we post this to the list??? i didn't wanna do that without your >> >> consent.?? it's just that maybe we can get someone else besides >> the two >> >> of us to pipe in on the conversation. >> >> > >> >> > >> >> > >> >> > ---------------------------- Original Message >> >> ---------------------------- >> >> > Subject: re Burlington >> >> > >> >> >> From: "Sennet Williams" <sennetwilliams@yahoo.com> >> >> > Date: Thu, May 16, 2019 8:54 pm >> >> > To: "rbj@audioimagination.com" <rbj@audioimagination.com> >> >> > >> >> >> -------------------------------------------------------------------------- >> >> > >> >> >> well, I have no idea what IRV system Burlington used. >> >> > >> >> > it's the same IRV as in every other governmental RCV election except >> >> we had 5 ranking levels and 5 candidates.?? so no one was >> >> "disenfranchised".?? you could have ranked the candidates in opposite >> >> order of their expost facto popularity, and you would still be able to >> >> weigh in on the IRV final round that actually selects the mayor. >> >> > >> >> > here is an analysis of what went >> >> wrong:??https://rangevoting.org/Burlington.html >> >> > >> >> > here's >> >> >> another:??http://bolson.org/~bolson/2009/20090303_burlington_vt_mayor.html >> >> > >> >> > essentially, we had 4 strong candidates going in.?? 3 were all >> >> plausible winners.?? the GOP candidate had the Plurality, the Prog >> >> candidate wonthe IRV, and the Dem candidate was the pairwise >> >> champion.?? the Dem would have beaten **any** other candidate in the >> >> IRV final round had he been able to advance to the final round.?? that >> >> IRV eliminated him in the semi-final round was the execution of this >> >> inherent flaw of IRV. >> >> > >> >> > >> >> >> The problem we have had in SF, Berkeley and Oakland is that each >> >> voter can only select three candidates, and the number of exhausted >> >> ballots exceeded the winning margin in at least several elections. >> >> > >> >> > yes, that's a known problem with **any** RCV if the number of >> >> candidates on the ballotexceeds the number of ranking levels.?? you >> >> need more ranking levels than 3 and you need stronger (or stricter) >> >> ballot access requirements so that fewer candidates get on the ballot >> >> and only those that are plausible winners.?? i think 5 levels is >> >> enough, and the number of signatures on apetition needed to get on the >> >> ballot can be adjusted by law in response to the usual number of >> >> candidates that make it onto the ballot.?? if there are consistently >> >> more names than ranking levels, the legislative body has the >> >> information and the authority necessary to increase the number >> >> ofrequired signatures to have candidate access to the ballot. >> >> > >> >> > >> >> >> Most clearly in the Kaplan/Quan/Perata mayor's contest (Oakland''s >> >> 1st IRV election)?? There were also six "minor" candidates.?? Kaplan >> >> was?? almost surely the most preferred, but Quan gamed the system by >> >> mortgaging her house and spending a lot asking casual voters to"make >> >> me 2nd.?? The winning margin over Kaplan was very narrow but the >> number >> >> of exhausted ballots was very large because most of the minor >> >> candidates were black while none of the big three were.???? A lot of >> >> people blamed the IRV system for electing Quan, who was >> >> basicallyincompetent, but there has been no serious attempt to >> repeal IRV. >> >> > >> >> > >> >> > >> >> > Ranked-Choice Voting will not stop bad politicians that are good >> >> salespersons from winning office.?? But it is intended to stop spoiler >> >> candidates from preventing the candidate with the actual popular >> >> supportfrom winning. >> >> > >> >> > >> >> >> When CA gets statewide IRV, we would presumably Maine's system and >> >> all counties will be given new equipment so all candidates can be >> ranked. >> >> > >> >> > In Burlington we didn't need new equipment.?? just new >> software.?? the >> >> optical-scan machines were the same machines, but they had to >> >> beprogrammed slightly differently. >> >> > >> >> > Sennet, do you understand what the difference is between IRV and >> >> Condorcet??? What it is that we on the list bitch about regarding IRV. >> >> > >> >> > Our issue is not that we don't like RCV, we **want** Ranked-Choice >> >> Voting, we just want the rules reformed so thatthe pairwise champion >> >> is always elected.?? IRV will do that *most* of the time, but it does >> >> not always do that.?? and like the Electoral College, when IRV >> fails to >> >> elect who we all know should have been elected, it never brings >> >> legitimacy to the election.?? failure to elect the pairwisechampion >> >> will only harm voting system reform. >> >> > >> >> > >> >> > >> >> > -- >> >> > >> >> > r b-j?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ??rbj@audioimagination.com >> >> > >> >> > "Imagination is more important than knowledge." >> >> > >> >> ?? >> >> ?? >> >> ?? >> >> >> ---- >> Election-Methods mailing list - seehttps://electorama.com/em for list info > > <http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient> > Virus-free. www.avg.com > <http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient> > > > <#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2> > > ---- > Election-Methods mailing list - see https://electorama.com/em for list info

Nearly all voters can vote against a candidate, be unwilling to accept that candidate as a winner, and the candidate can still be a Cordorcet winner. This becomes more possible as the candidate set enlarges. It is easiest to understand this if all voters truncate, effectively equal-ranking all but one bottom. Simple example: two voters prefer a single candidate over all others. All other voters are divided, voting only to prefer their own favorite, each one different. So in every pairwise race, the candidate wins. Condorcet is an interesting criterion, but is far from the whole of what is desirable. In traditional voting, before modern elections, no decision was made unless a majority supported it. Plurality elections discarded that principle in the interest of efficiency, which I have been pointing out is fascist. If a Condorcet method tests approval of results, and requires a majority approving to complete the election, it can avoid the problem, but at the risk of failing to complete. It has been common to accept the need for "runoff elections." That could be drastically improved by using advanced election methods for both elections, and using much more intelligent methods of selecting candidates for a runoff. With advanced methods, a Condorcet winner in the primary would always advance to the runoff and the issue would be the other one or two. And then, again, having approval indication, what should happen if no candidate gains majority approval in the runoff? I classic democracy, they simply kept on voting until they had majority approval of an outcome. On 5/20/2019 2:52 PM, Richard Lung wrote: > "But it is only Condorcet that elects the candidate that is explicitly > preferred by voters over every other candidate." > > I wonder tho, whether that satisfies the requiremant for one candidate > (of their number) to be prefered over a whole range of candidates? >

---------------------------- Original Message ---------------------------- Subject: Re: [EM] re Burlington From: "Abd ul-Rahman Lomax" <abd@lomaxdesign.com> Date: Thu, May 30, 2019 4:01 pm To: election-methods@lists.electorama.com -------------------------------------------------------------------------- > Nearly all voters can vote against a candidate, be unwilling to accept > that candidate as a winner, and the candidate can still be a Condorcet > winner. This becomes more possible as the candidate set enlarges. > > It is easiest to understand this if all voters truncate, effectively > equal-ranking all but one bottom. > > Simple example: two voters prefer a single candidate over all others. > All other voters are divided, voting only to prefer their own favorite, > each one different. So in every pairwise race, the candidate wins. > Condorcet is an interesting criterion, but is far from the whole of what > is desirable. but, as a general rule for real elections in which the voter participation is in the thousands (or more), if the Condorcet Winner exists and they elect someone else, how do they justify that to voters as a body?  How do you tell voters that more of them voted for A over B than the number that voted for B over A and yet B gets elected?  Assuming a CW exists, how possibly can the election of B be considered more democratic? > In traditional voting, before modern elections, no decision was made > unless a majority supported it. Plurality elections discarded that > principle in the interest of efficiency, which I have been pointing out > is fascist. If a Condorcet method tests approval of results, and > requires a majority approving to complete the election, it can avoid the > problem, but at the risk of failing to complete. It has been common to > accept the need for "runoff elections." the commonly cited problem with runoff elections that occur at a later date is that, even in communities with relatively large voter participation, the participation for runoffs is usually around 50% of the number that voted in the original election. one of the selling points of RCV (over FPTP with top-two runoff if necessary) is that the election is **decisive** on election day.  people can give their contingency vote on election day as to whom they would vote for in a runoff if their favorite candidate is eliminated.  i believe this is what motivated the STV procedure of instant runoff voting. > That could be drastically > improved by using advanced election methods for both elections, and > using much more intelligent methods of selecting candidates for a > runoff. With advanced methods, a Condorcet winner in the primary would > always advance to the runoff and the issue would be the other one or > two. And then, again, having approval indication, what should happen if > no candidate gains majority approval in the runoff? I classic democracy, > they simply kept on voting until they had majority approval of an outcome. > > On 5/20/2019 2:52 PM, Richard Lung wrote: >> "But it is only Condorcet that elects the candidate that is explicitly >> preferred by voters over every other candidate." >> >> I wonder tho, whether that satisfies the requiremant for one candidate >> (of their number) to be prefered over a whole range of candidates? >> -- r b-j                         rbj@audioimagination.com "Imagination is more important than knowledge."        

> It is easiest to understand this if all voters truncate, effectively equal-ranking all but one bottom. I guess Condorcet and most methods resemble plurality if voters so want. In most methods they can just bullet vote and leave the rest of the information out. > In traditional voting, before modern elections, no decision was made unless a majority supported it. One could try to reach majority by making voting mandatory. Or if majority of those that vote is enough, then one could make filling the ballot paper appropriately mandatory. This covers both the first round and the runoff. How abut trying to solve this problem in Condorcet by introducing a "status quo" candidate in the election? If that candidate wins, then the election is declared inconclusive. Or alternatively people have decided to keep the status quo. If needed, one could arrange a second round where people would know better which candidates to rank. One interesting question here is, what is the default position of the "status quo" candidate. It could be shared last, as with other candidates, but one could also think it would be "lonely last" behind others, or first position. Juho P.S. I wonder also if there would be need for a second additional candidate. That threshold could be e.g. "no way worse than this, I'm serious about keeping the status quo" :-). Candidates that lose to that threshold could e.g. be excluded from the second round (if "status quo" won the first round). > On 31 May 2019, at 02:01, Abd ul-Rahman Lomax <abd@lomaxdesign.com> wrote: > > Nearly all voters can vote against a candidate, be unwilling to accept that candidate as a winner, and the candidate can still be a Cordorcet winner. This becomes more possible as the candidate set enlarges. > > It is easiest to understand this if all voters truncate, effectively equal-ranking all but one bottom. > > Simple example: two voters prefer a single candidate over all others. All other voters are divided, voting only to prefer their own favorite, each one different. So in every pairwise race, the candidate wins. Condorcet is an interesting criterion, but is far from the whole of what is desirable. > > In traditional voting, before modern elections, no decision was made unless a majority supported it. Plurality elections discarded that principle in the interest of efficiency, which I have been pointing out is fascist. If a Condorcet method tests approval of results, and requires a majority approving to complete the election, it can avoid the problem, but at the risk of failing to complete. It has been common to accept the need for "runoff elections." That could be drastically improved by using advanced election methods for both elections, and using much more intelligent methods of selecting candidates for a runoff. With advanced methods, a Condorcet winner in the primary would always advance to the runoff and the issue would be the other one or two. And then, again, having approval indication, what should happen if no candidate gains majority approval in the runoff? I classic democracy, they simply kept on voting until they had majority approval of an outcome. > > On 5/20/2019 2:52 PM, Richard Lung wrote: >> "But it is only Condorcet that elects the candidate that is explicitly preferred by voters over every other candidate." >> >> I wonder tho, whether that satisfies the requiremant for one candidate (of their number) to be prefered over a whole range of candidates? >> > ---- > Election-Methods mailing list - see https://electorama.com/em for list info

On 31/05/2019 08.39, Juho Laatu wrote: >> It is easiest to understand this if all voters truncate, >> effectively equal-ranking all but one bottom. > > I guess Condorcet and most methods resemble plurality if voters so > want. In most methods they can just bullet vote and leave the rest of > the information out. > >> In traditional voting, before modern elections, no decision was >> made unless a majority supported it. > > One could try to reach majority by making voting mandatory. Or if > majority of those that vote is enough, then one could make filling > the ballot paper appropriately mandatory. This covers both the first > round and the runoff. > > How abut trying to solve this problem in Condorcet by introducing a > "status quo" candidate in the election? If that candidate wins, then > the election is declared inconclusive. Or alternatively people have > decided to keep the status quo. If needed, one could arrange a second > round where people would know better which candidates to rank. > > One interesting question here is, what is the default position of the > "status quo" candidate. It could be shared last, as with other > candidates, but one could also think it would be "lonely last" behind > others, or first position. Combining Condorcet with the sort of repeated runoff Abd suggests is pretty easy. Just repeatedly eliminate everybody who falls outside of the Smith set until only one candidate remains; he's the winner. Preventing such a method from going on forever is a different matter, though. Perhaps Heitzig's consensus system could be used to force random ballot after a certain point, but it isn't clear to me just how you'd do that.