Hi Mike, May I suggest that you also include a 3-slot ballot option? I.e., Preferred, Acceptable, Reject. You could call it a Fallback Approval ballot if you like.
Many methods (e.g., most Condorcet methods, ER-Bucklin) that don't meet the Participation criterion will do so when restricted to 3-slots. It would be interesting to compare behavior with that level of compression. Ted On 06 Jan 2012 13:56:54 -0800, MIKE OSSIPOFF wrote: > > I've long advocated that you can't adequately discuss the relative merits or > desirability of > voting systems without actually using them. ...without actually trying them > out. For that, > it's absolutely essential to do polling, simulated political elections, using > the methods that are > proposed at EM. > > You don't know the problems of methods that you consider best, until you use > them in > an election, even if a simulated election. > > Therefore, I propose a simulated presidential election. Mainly because we > don't > know who > the 2012 nominations will be yet, I suggest that the voting be by party, > instead of by > candidate. In some ways, that's more meaningful anyway, because policy > platforms are, > or should be, the basis of political voting. > > Some have claimed that we should do polling at external websites, automated > websites. The > main problem with that is flexibility: EM polls have nearly always included > balloting by Approval, > Score Voting, and ranking. And they always should, because all of those > balloting modes are > used by some of the various methods proposed on EM. My poll includes all three > of those > balloting modes...three separate ballots: Approval, Score, and rank. > > Another problem with automated polling websites is ballot-stuffing. > Even though polling websites usually register voters by their > e-mail, that only reduces, but doesn't eliminate the possibility of > ballot-stuffing. Of course that > problem isn't as important in a poll whose only purpose is to demonstrate what > it's like to use > the various voting systems. But, arguably, it still matters, for the purpose > of > such polls, that the > observed result reliably reflect the 1-per-voter ballots. > > This poll could be criticized because EM's membership is international, and > I'm > proposing a > simulated U.S. election. I invite non-U.S. members to vote in this poll, > because its purpose is > merely to demonstrate the use of the proposed voting systems. > > If poll-participants identify themselves, in parenthesis as "international" or > "U.S.", then separate > election results can be determined, one of which would indicate what kind of a > party is > the EM international winner, and the other of which would indicate which kind > of party would > win in the U.S. if EM members are typical. > > ...And EM members are more typical than some might believe, in terms of their > sincere > preferences. I'd suggest that EM members differ from the general public mostly > in that they > aren't Republocrat lesser-of-2-evils voters. Even if some EM members actually > prefer > the Republicans or Democrats, none will favor one of those parties only as a > lesser-evil. > > Strategy? I suggest that any strategy used in this simulated election be > appropriate to the > EM electorate. If you perceive any difference between the EM electorate and > the > general > population, then base your strategy on the EM electorate. It makes a poll more > realistic > if voting is based on the conditions in the poll. > > Should Score voting be sincere, or should it be however you'd vote it in an > actual public > political election? I suggest the latter. > > Sure, with an Approval balloting, it could be argued that there's no need for > Approval > strategy in Score voting, so the Score voting should be sincere, regardless of > whether > you'd rate sincerely in an actual election. I and others have made that > suggestion in > previous EM polls. > > But I don't think that's best in this poll. The purpose of this poll is to try > out the various > methods, not to determine the sincere Score winner among the EM electorate. So > I > suggest voting the Score ballot exactly as you would if it were an actual > public political > election, in which Score voting were the only kind in use. > > Unless Warren argues for suggesting sincere ratings on the Score ballot, I > suggest > voting the same ratings you'd vote in an actual public political Score > election. Because > we want to simulate an actual election. > > As you know, I advocate, as options in an Approval balloting, the following > ways of > voting: > > Approval, MTA, MCA, ABucklin, AOC, MTAOC, MCAOC, and AOCBucklin. > > I'll define these ways of voting in a subsequent posting. > > But I'll briefly outline their definitions here: > > You know what Approval, MTA and MCA are. > > AOC is Approval, with the option to make some approvals conditional upon > mutuality, as defined > by the MTAOC pseudocode program that I posted here. > > MTAOC and MCAOC are MTA and MCA with that conditionality option. > > AOCBucklin is ABucklin with that option at each rank position. > > When there are ballots using the ABucklin &/or AOCBucklin option, the election > is equivalent to > an ABucklin election, and is counted as such. Of course an Approval ballot > counts as an ABucklin > ballot that only gives first preferences. > > MCA and MTA are counted in the obvious way that I previously described, > consistent with and > compatible with Approval and ABucklin. > > All of the above remains true when ballots also use the conditionality option. > > For AOCBucklin, all votes that have been assigned to a candidate, other than > 1st preference votes, > come under the term "middle ratings", for the purposes of MTAOC > conditionality. > > For AOCBucklin, the conditionality calculations must be done anew after each > AOCBucklin vote-assignment > stage. That's because the new vote assignments change the middle ratings > counts that the MTAOC > conditionality calculation uses. > > Voting instructions: > > For the Approval election: > > You can vote a ballot by: > > Approval, MTA, MCA, ABucklin, AOC, MTAOC, MCAOC, or AOCBucklin. > > If you choose an Approval ballot, approve whichever candidate(s) you choose > to. > You have the option of > designating any of those approvals as conditional. That indicates that you > don't want that approval to be > usable to defeat your approved candidates. You give it conditional upon its > being reciprocated as defined > in the MTAOC program pseudocode that I posted. (More about the conditionality > option later in this posting). > > If you choose an MTA or MCA ballot, then you rate some candidates then > indicate > that you rate > some candidates "top" and some candidates "middle". As with Approval, you can > indicate that some > of your middle ratings are conditional, meaning that you don't want them to be > usable to defeat your > coalition-suitable candidates (defined later in this posting). And, as said > before, that guarantee is carried > out by making that middle rating conditional upon reciprocity as defined in > the > MTAOC program pseudocode. > > If you choose an ABucklin ballot, then rank any number of candidates in order > of preference. At any rank > position, you can indicate, for any candidate, that your vote for hir is > conditional, as defined above. > > Every method with more than two slots can benefit by AERLO, the Automatic > Equal > Ranking Line Option. > > To use AERLO in MTA or MCA, list your middle-rated candidates vertically in > order of preference, with "AERLO" written just > below the lowest-ranked one that you want to be protected by the AERLO option. > > In ABucklin, write AERLO just below the lowest rank position that you want to > protect with AERLO. > > What it means when you have AERLO in your ranking, or your middle MTA or MCA > ratings: > > If none of your above-AERLO candidates wins, then they are all moved to > top-rating or top-ranking. Another > count is conducted after that raising-to-top has done on each such ballot. > That > winner of that 2nd count > wins the election. > > For example, an AOCBucklin ballot with AERLO might look like this: > > 1. Candidate A > 2. Candidate B > AERLO > 3. Candidate C > 4. Candidate D (conditional) > > The conditionality option, should you invoke it, requires that a middle rating > be mutual as defined > in the MTAOC program pseudocode. That program refers to candidates whom you > designate > "coalition-suitable". You don't have to actually make those designations. > > The default "coalition-suitable" designation is: Your above-AERLO candidates > are coalition-suitable. > If you don't use AERLO, then your top-rated or top-ranked candidates are > coalition-suitable. > > That's the default. > > But, if you want to, you could specify that you only want your _initially_ > top-rated candidates to be > coalition-suitable. > > Or you could even designate particular candidates as candidate suitable, if > you > choose to. Doing so means > that you aren't using the default assumption. > > But you needn't bother with that, because there is a useful and practical > default assumption, stated above. > > For the Score election: > > Rate the parties as you would in an actual public election. If you strategize, > do so with respect to the > actual EM electorate. > > (unless Warren asks that people rate sincerely, with strategy only in the > Approval election rather than in the > Score election) > > For the ranking election: > > Best to rank sincerely. Anyone can count the rankings by any method they > choose. I'll count them > by MMPO with AERLO. No need to count them by AOCBucklin or ABucklin, since > that > way of voting is > included as an option in the Approval election. > > Also, because the rankings election is intended to be method-nonspecific, it's > best to count it be methods > that don't require much other than just a ranking. Count methods with drastic > voting strategy aren't > desirable count methods, of course. > > I suggest that the AERLO option should be available for the rank election, > because AERLO is useful in pretty > much every rank method that allows equal ranking. I consider MMPO with AERLO > to > be a good method, and, if there > is participation in this mock election, I'll do an MMPO with AERLO count for > the ranking election. > > Voter's Choice: > > Though this is not the purpose of this mock election, it's possible to > determine an overall winner, by Voter's Choice: > > When you vote, designate a method. Your designated method could be Approval, > or > Score, or any rank-count. > > After the counts are completed, each political party receives a score equal to > the sum of the numbers of people > designating the methods by which that party won. > > The winning party is the one with the highest such score. > > (MTA, MCA, ABuckliln, AOC, MTAOC, MCAOC and AOCBucklin don't count as separate > election methods, as they are only > voting options for the approval election). > > Nominations: > > Every election should have a nomination period. I suggest a one-week > nomination > period. > > I don't know how you feel about having a campaign period, because maybe this > list shouldn't > have political advocacy. If people feel that there should be no campaigning, > no > discussion of the > parties' relative merits, then that's fine. > > If people feel that such discussion is appropriate for an election, then there > could be a week set aside > for it, after the nominations week. > > I suggest that nominations can be made any time starting right now, and > continuing till 0 hours, 1 minute > GMT (UT), January 15th. > > I'll just start by making a few obvious nominations. Nominating a party > doesn't > mean that the nominator > likes it. It might just be that (as in the case of some that I nominate) those > parties are just felt to be > ones that would be found in an actual election, and are nominated for that > reason only. > > Of course no one needs to rank or rate all of the parties nominated. > > With that understanding, I nominate the following parties, listed in > alphabetical order, to avoid > the appearance of favoritism: > > Boston Tea Party (not to be confused with the Republican-like "Tea Party > Movement") > Democrats (moderate) (example: Mondale) > Democrats (Republican-like) (example: Lieberman) > Democrats (relatively progressive) (example: Kuccinich) > Greens/Green Party USA (G/GPUSA) (The original U.S. Greens) > Green Party US (GPUS) (The replacement Greens) > Libertarians (as defined by the Libertarian platform on the Internet) > Republicans (Moderate) > Republicans (More Republican) > Socialist Party USA (SPUSA) > > That seems to roughly span the political spectrum among the U.S. political > parties. > > Of course you might want to look up those parties' platforms on the Internet. > > Also useful, though biased, might be the "directory of U.S. political parties" > on the Internet. > > (I'm not quite sure how its name is worded, but the above name will probably > find it at a > search engine). > > By the way, what if I throw a party and no one comes? Or propose an election > and no > one participates? I feel that mock elections are much needed at a voting > system > discussing > mailing list. I'm doing my part by proposing this mock election. That's so > regardless of whether > anyone else thinks there should be a mock election, and regardless of whether > anyone > participates. My purpose is merely to propose the election, make it available. > Having done so, > I've done my part. > > Mike Ossipoff > > > ---- > Election-Methods mailing list - see http://electorama.com/em for list info -- araucaria dot araucana at gmail dot com ---- Election-Methods mailing list - see http://electorama.com/em for list info
