<html><head></head><body><div class="ydp8f615ceyahoo-style-wrap" style="font-family:Helvetica Neue, Helvetica, Arial, sans-serif;font-size:16px;"><div></div>
        <div>Hi Chris,</div><div><br></div><div><div><div><blockquote type="cite" style="margin: 0px; padding: 8px; color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;">>>I don't think it's ideal if burying X under Y (both disapproved) can only backfire when Y is made the CW.<div>>></div></blockquote><span style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;">>Why is that?  </span></div><div><span style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;"><br></span></div><div><span style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;">Because I think if voters decide to attempt to prevent another candidate from being CW, via insincerity, there should be risks to doing that. Of course there is already some risk. But if you "knew" that a given candidate had no chance of being CW then there would be nothing to lose in using that candidate in a burial strategy.</span></div><div><span style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;"><br></span></div><div><span style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;">>The post-election complaint (by any of the voters) would be .. what?</span></div><div><br></div><div>For either a successful burial strategy, or one that backfires and elects an arbitrary candidate, I think the possible complaints are clear. Maybe someone would argue that a backfiring strategy proves the method's incentives are just fine. But that wouldn't be how I see it. I think if, in actual practice, it ever happens that voters calculate that a strategy is worthwhile, and it completely backfires to the point that everyone would like the results discarded, then that method will probably get repealed.</div><div><br></div><div>><br clear="none" style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;"><span style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;">>If you don't allow voters to rank among their unapproved candidates then arguably you are not even trying to elect the sincere CW.</span><br clear="none" style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;"><span style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;">>Instead you are just modifying Approval to make it a lot more Condorcet-ish.  </span></div><div><br></div><div>Not an unfair statement. If you require voters to have that much expressiveness then you can't use implicit.</div><div><br></div><div>To me, the motivation for three-slot C//A(implicit) is partly about burial, partly about method simplicity, partly about ballot simplicity. C//A(explicit) retains 1 of 3. (Arguably slightly less for the Smith version.) Possibly it has its own merits, but they will largely be different ones.</div><div><br clear="none" style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;">><br clear="none" style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;"><span style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;">>A lot of voters like relatively expressive ballots. I think that is one of the reasons why Approval seems to be a lot less popular than IRV.</span><br clear="none" style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;"></div></div><div><span style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;"><br></span></div><div><span style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;">I have no *inherent* complaints about the ballot format of explicit approval plus full ranking.</span></div><div><span style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;"><br></span></div><div><span style="color: rgb(38, 40, 42); font-family: Helvetica Neue, Helvetica, Arial, sans-serif;">Kevin</span></div><br></div><div><br></div><div><br></div><div><br></div>
        
        </div><div id="ydpb6a32dd9yahoo_quoted_0445501564" class="ydpb6a32dd9yahoo_quoted">
            <div style="font-family:'Helvetica Neue', Helvetica, Arial, sans-serif;font-size:13px;color:#26282a;">
                
                <div>
                    Le jeudi 6 juin 2019 à 21:03:19 UTC−5, C.Benham <cbenham@adam.com.au> a écrit :
                </div>
                <div><br></div>
                <div><br></div>
                <div><div id="ydpb6a32dd9yiv8583323751"><div>
    <p>Kevin,<br clear="none">
      </p><blockquote type="cite">Specifically should "positional dominance"
        have the same meaning whether or not the method has approval in
        it?</blockquote>
      <br clear="none">
      If the voters all choose to approve all the candidates they rank,
      then yes.  (For a while I was wrongly assuming that Forest's
      suggested<br clear="none">
      default approval was for all ranked-above-bottom candidates, but
      then I noticed that he specified that it was only for top voted
      candidates).<br clear="none">
      <br clear="none">
      One of my tired examples:<br clear="none">
      <br clear="none">
      25: A>B<br clear="none">
      26: B>C<br clear="none">
      23: C>A<br clear="none">
      26: C<br clear="none">
      <br clear="none">
      Assuming all the ranked candidates are approved, C is by far the
      most approved and the most top-voted candidate. <br clear="none">
      Normal Winning Votes (and your idea 2 in this example) elect B.<br clear="none">
      <br clear="none">
      <blockquote type="cite">I will go easy on these methods over
        failing MD, because it happens when some of the majority don't
        approve their common candidate.</blockquote>
      <br clear="none">
      For me this this type of ballot avoids the Minimal Defense versus
      Chicken Dilemma dilemma, rendering those criteria inapplicable.<br clear="none">
    
    <p>48: A<br clear="none">
      27: B>C<br clear="none">
      25: C<br clear="none">
      <br clear="none">
      The problem has been that we don't know whether the B>C voters
      are thinking "I am ranking C because above all I don't want that
      evil A<br clear="none">
      to win" or  "My C>A preference isn't all that strong, and I
      think that my favourite could well be the sincere CW, and if  C's
      supporters rank<br clear="none">
      B above A then B has a good chance to win. But if they if they
      create a cycle by truncating I'm not having them steal it".<br clear="none">
      <br clear="none">
      With the voters able to express explicit approval we no longer
      have to guess which it is.<br clear="none">
      <br clear="none">
      </p><blockquote type="cite"> I don't think it's ideal if burying X
        under Y (both disapproved) can only backfire when Y is made the
        CW.
        <div><br clear="none">
        </div>
      </blockquote>
      Why is that?  The post-election complaint (by any of the voters)
      would be .. what?<br clear="none">
      <br clear="none">
      If you don't allow voters to rank among their unapproved
      candidates then arguably you are not even trying to elect the
      sincere CW.<br clear="none">
      Instead you are just modifying Approval to make it a lot more
      Condorcet-ish.  <br clear="none">
      <br clear="none">
      A lot of voters like relatively expressive ballots. I think that
      is one of the reasons why Approval seems to be a lot less popular
      than IRV.<br clear="none">
      <br clear="none">
      Chris Benham<br clear="none">
      <br clear="none">
    
    <div class="ydpb6a32dd9yiv8583323751yqt8593146214" id="ydpb6a32dd9yiv8583323751yqt96283"><div class="ydpb6a32dd9yiv8583323751moz-cite-prefix">On 6/06/2019 5:34 pm, Kevin Venzke
      wrote:<br clear="none">
    </div>
    <blockquote type="cite">
      </blockquote></div></div><div class="ydpb6a32dd9yiv8583323751yqt8593146214" id="ydpb6a32dd9yiv8583323751yqt32286"><div><div class="ydpb6a32dd9yiv8583323751ydpe4c7db39yahoo-style-wrap" style="font-family:Helvetica Neue, Helvetica, Arial, sans-serif;font-size:16px;">
        <div>Hi Chris,</div>
        <div><br clear="none">
        </div>
        <div>I've been short on time so I don't actually have much
          thought on any of the methods, even my own.</div>
        <div><br clear="none">
        </div>
        <div>I suppose Idea 2 is the same as Schwartz-limited MinMax(WV)
          if nobody submits disapproved rankings. I'm not sure if it
          makes sense to reject the method over that. Specifically
          should "positional dominance" have the same meaning whether or
          not the method has approval in it? As a comparison, I will go
          easy on these methods over failing MD, because it happens when
          some of the majority don't approve their common candidate.</div>
        <div><br clear="none">
        </div>
        <div>I would have liked to simplify Idea 2, but actually
          Forest's eventual proposal wasn't all that simple either. As I
          wrote, if you add "elect a CW if there is one" it can become
          much simpler, so that it isn't really distinct from Idea 1. I
          actually tried pretty hard to present three "Ideas" in that
          post, but kept having that problem.</div>
        <div><br clear="none">
        </div>
        <div>I posted those ideas because I thought Forest posed an
          interesting challenge, and I thought I perceived that he was
          trying to fix a problem with CD. That said, I am not a fan of
          Smith//Approval(explicit). If all these methods are basically
          the same then I probably won't end up liking any of them. I
          don't think it's ideal if burying X under Y (both disapproved)
          can only backfire when Y is made the CW.</div>
        <div><br clear="none">
        </div>
        <div>Kevin</div>
        <div><br clear="none">
        </div>
        <div><br clear="none">
        </div>
      </div>
      <div class="ydpb6a32dd9yiv8583323751ydp50d5cd69yahoo_quoted" id="ydpb6a32dd9yiv8583323751ydp50d5cd69yahoo_quoted_0454840046">
        <div style="font-family:'Helvetica Neue', Helvetica, Arial, sans-serif;font-size:13px;color:#26282a;">
          <div> Le mercredi 5 juin 2019 à 21:26:23 UTC−5, C.Benham
            <a shape="rect" class="ydpb6a32dd9yiv8583323751moz-txt-link-rfc2396E" href="mailto:cbenham@adam.com.au" rel="nofollow" target="_blank"><cbenham@adam.com.au></a> a écrit : </div>
          <div><br clear="none">
          </div>
          <div>Kevin,<br clear="none">
          </div>
          <div>
            <div id="ydpb6a32dd9yiv8583323751ydp50d5cd69yiv9085021920">
              <div>
                <p>I didn't comment earlier on your "idea 2".  <br clear="none">
                  <br clear="none">
                  If there no "disapproved rankings" (i.e. if the voters
                  all approve the candidates they rank above bottom),<br clear="none">
                  then your suggested method is simply normal  Winning
                  Votes, which I don't like because the winner can<br clear="none">
                  be uncovered and positionally dominant or
                  pairwise-beaten and positionally dominated by a single
                  other<br clear="none">
                  candidate.<br clear="none">
                  <br clear="none">
                  On top of that I don't think it really fills the bill
                  as "simple".  Approval Margins (using Sort or
                  Smith//MinMax<br clear="none">
                  or equivalent or almost equivalent algorithm) would be
                  no more complex and in my opinion would be better.<br clear="none">
                  <br clear="none">
                  I would also prefer the still more simple
                  Smith//Approval.<br clear="none">
                  <br clear="none">
                  What did you think of my suggestion for a way to
                  implement your idea 1?  </p>
                <div class="ydpb6a32dd9yiv8583323751ydp50d5cd69yiv9085021920yqt3873327189" id="ydpb6a32dd9yiv8583323751ydp50d5cd69yiv9085021920yqtfd25173"><br clear="none">
                  Chris <br clear="none">
                </div>
                <div class="ydpb6a32dd9yiv8583323751ydp50d5cd69yiv9085021920yqt3873327189" id="ydpb6a32dd9yiv8583323751ydp50d5cd69yiv9085021920yqtfd71007">
                  <div class="ydpb6a32dd9yiv8583323751ydp50d5cd69yiv9085021920moz-forward-container"><br clear="none">
                    <br clear="none">
                    <br clear="none">
                    <blockquote type="cite">
                      <p><b>Kevin Venzke</b> <a shape="rect" title="[EM] What are some simple methods that                           accomplish the following conditions?" href="mailto:election-methods%40lists.electorama.com?Subject=Re%3A%20%5BEM%5D%20What%20are%20some%20simple%20methods%20that%20accomplish%20the%20following%0A%20conditions%3F&In-Reply-To=%3C1931864740.14928463.1559418507456%40mail.yahoo.com%3E" rel="nofollow" target="_blank">stepjak at yahoo.fr </a><br clear="none">
                        Sat Jun 1 12:48:27 PDT 2019 </p>
                      <p><br clear="none">
                        Hi Forest,<br clear="none">
                        <br clear="none">
                        I had two ideas.<br clear="none">
                        <br clear="none">
                        Idea 1:<br clear="none">
                        1. If there is a CW using all rankings, elect
                        the CW.<br clear="none">
                        2. Otherwise flatten/discard all disapproved
                        rankings.<br clear="none">
                        3. Use any method that would elect C in scenario
                        2. (Approval, Bucklin, MinMax(WV).)<br clear="none">
                        <br clear="none">
                        So scenario 1 has no CW. The disapproved C>A
                        rankings are dropped. A wins any method.<br clear="none">
                        In scenario 2 there is no CW but nothing is
                        dropped, so use a method that picks C.<br clear="none">
                        In both versions of scenario 3 there is a CW, B.<br clear="none">
                        <br clear="none">
                        If step 3 is Approval then of course step 2 is
                        unnecessary.<br clear="none">
                        <br clear="none">
                        In place of step 1 you could find and apply the
                        majority-strength solid coalitions (using all
                        rankings)<br clear="none">
                        to disqualify A, instead of acting based on B
                        being a CW. I'm not sure if there's another
                        elegant way<br clear="none">
                        to identify the majority coalition.<br clear="none">
                        <br clear="none">
                        Idea 2:<br clear="none">
                        1. Using all rankings, find the strength of
                        everyone's worst WV defeat. (A CW scores 0.)<br clear="none">
                        2. Say that candidate X has a "double beatpath"
                        to Y if X has a standard beatpath to Y
                        regardless<br clear="none">
                        of whether the disapproved rankings are counted.
                        (I don't know if it needs to be the *same*
                        beatpath,<br clear="none">
                        but it shouldn't come into play with these
                        scenarios.)<br clear="none">
                        3. Disqualify from winning any candidate who is
                        not in the Schwartz set calculated using double<br clear="none">
                        beatpaths. In other words, for every candidate Y
                        where there exists a candidate X such that X has
                        a<br clear="none">
                        double beatpath to Y and Y does not have a
                        double beatpath to X, then Y is disqualified.<br clear="none">
                        4. Elect the remaining candidate with the
                        mildest WV defeat calculated earlier.<br clear="none">
                        <br clear="none">
                        So in scenario 1, A always has a beatpath to the
                        other candidates, no matter whether disapproved<br clear="none">
                        rankings are counted. The other candidates only
                        have a beatpath to A when the C>A win exists.
                        So<br clear="none">
                        A has a double beatpath to B and C, and they
                        have no path butt. This leaves A as the only
                        candidate<br clear="none">
                        not disqualified.<br clear="none">
                        <br clear="none">
                        In scenario 2, the defeat scores from weakest to
                        strongest are B>C, A>B, C>A. Every
                        candidate has<br clear="none">
                        a beatpath to every other candidate no matter
                        whether the (nonexistent) disapproved rankings
                        are<br clear="none">
                        counted. So no candidate is disqualified. C has
                        the best defeat score and wins.<br clear="none">
                        <br clear="none">
                        In scenario 3, the first version: B has no
                        losses. C's loss to B is weaker than both of A's
                        losses. B<br clear="none">
                        beats C pairwise no matter what, so B has a
                        double beatpath to C. However C has no such
                        beatpath<br clear="none">
                        to A, nor has A one to B, nor has B one to A.
                        The resulting Schwartz set disqualifies only C.
                        (C needs<br clear="none">
                        to return B's double beatpath but can't, and
                        neither A nor B has a double beatpath to the
                        other.)<br clear="none">
                        Between A and B, B's score (as CW) is 0, so he
                        wins.  <br clear="none">
                        <br clear="none">
                        Scenario 3, second version: B again has no
                        losses, and also has double beatpaths to both of
                        A and<br clear="none">
                        C, neither of whom have double beatpaths butt.
                        So A and C are disqualified and B wins.<br clear="none">
                        <br clear="none">
                        I must note that this is actually a Condorcet
                        method, since a CW could never get disqualified
                        and<br clear="none">
                        would always have the best worst defeat. That
                        observation would simplify the explanation of<br clear="none">
                        scenario 3.<br clear="none">
                        <br clear="none">
                        I needed the defeat strength rule because I had
                        no way to give the win to B over A in scenario 3<br clear="none">
                        version 1. But I guess if it's a Condorcet rule
                        in any case, we can just add that as a rule, and
                        greatly<br clear="none">
                        simplify it to the point where it's going to
                        look very much like idea 1. I guess all my ideas
                        lead me to<br clear="none">
                        the same place with this question.<br clear="none">
                        <br clear="none">
                        Oh well, I think the ideas are interesting
                        enough to post.<br clear="none">
                        <br clear="none">
                        Kevin<br clear="none">
                        <br clear="none">
                        >Le jeudi 30 mai 2019 à 17:32:42 UTC−5,
                        Forest Simmons <fsimmons at pcc.edu> a
                        écrit : <br clear="none">
                        ><br clear="none">
                        >In the example profiles below 100 = P+Q+R,
                        and  50>P>Q>R>0.  One consequence of
                        these constraints is that in all three profiles
                        below the cycle >A>B>C>A will
                        obtain.<br clear="none">
                        ><br clear="none">
                        >I am interested in simple methods that
                        always ...<br clear="none">
                        ><br clear="none">
                        >(1) elect candidate A given the following
                        profile:<br clear="none">
                        ><br clear="none">
                        >P: A<br clear="none">
                        >Q: B>>C<br clear="none">
                        >R: C,<br clear="none">
                        >and <br clear="none">
                        >(2) elect candidate C given<br clear="none">
                        >P: A<br clear="none">
                        >Q: B>C>><br clear="none">
                        >R: C,<br clear="none">
                        >and <br clear="none">
                        >(3) elect candidate B given<br clear="none">
                        <br clear="none">
                        ><br clear="none">
                        >P: A<br clear="none">
                        >Q: B>>C  (or B>C)<br clear="none">
                        >R: C>>B. (or C>B)<br clear="none">
                        ><br clear="none">
                        >I have two such methods in mind, and I'll
                        tell you one of them below, but I don't want to
                        prejudice your creative efforts with too many
                        ideas.<br clear="none">
                        ><br clear="none">
                        >Here's the rationale for the requirements:<br clear="none">
                        ><br clear="none">
                        >Condition (1) is needed so that when the
                        sincere preferences are<br clear="none">
                        <br clear="none">
                        ><br clear="none">
                        >P: A<br clear="none">
                        >Q: B>C<br clear="none">
                        >R: C>B,<br clear="none">
                        >the B faction (by merely disapproving C
                        without truncation) can defend itself against a
                        "chicken" attack (truncation of B) from the C
                        faction.<br clear="none">
                        ><br clear="none">
                        >Condition (3) is needed so that when the C
                        faction realizes that the game of Chicken is not
                        going to work for them, the sincere CW is
                        elected.<br clear="none">
                        ><br clear="none">
                        >Condition (2) is needed so that when 
                        sincere preferences are<br clear="none">
                        <br clear="none">
                        ><br clear="none">
                        >P: A>C<br clear="none">
                        >Q: B>C<br clear="none">
                        >R: C>A,<br clear="none">
                        >then the C faction (by proactively
                        truncating A) can defend the CW against the A
                        faction's potential truncation attack.<br clear="none">
                        ><br clear="none">
                        >Like I said, I have a couple of fairly
                        simple methods in mind. The most obvious one is
                        Smith\\Approval where the voters have <br clear="none">
                        >control over their own approval cutoffs (as
                        opposed to implicit approval) with default
                        approval as top rank only. The other <br clear="none">
                        >method I have in mind is not quite as <br clear="none">
                        >simple, but it has the added advantage of
                        satisfying the FBC, while almost always electing
                        from Smith.<br clear="none">
                        <br clear="none">
                        <br clear="none">
                        <br clear="none">
                        <br clear="none">
                        <br clear="none">
                        <br clear="none">
                      </p>
                    </blockquote>
                    <i><br clear="none">
                    </i> </div>
                  <div id="ydpb6a32dd9yiv8583323751ydp50d5cd69yiv9085021920DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2"><br clear="none">
                    <table style="border-top:1px solid #D3D4DE;"><tbody><tr><td colspan="1" rowspan="1" style="width:55px;padding-top:13px;"><a shape="rect" href="http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient" rel="nofollow" target="_blank"><img src="https://ipmcdn.avast.com/images/icons/icon-envelope-tick-green-avg-v1.png" alt="" style="width:46px;min-height:29px;" width="46" height="29" data-inlineimagemanipulating="true"></a></td><td colspan="1" rowspan="1" style="width:470px;padding-top:12px;color:#41424e;font-size:13px;font-family:Arial, Helvetica, sans-serif;line-height:18px;">Virus-free.
                            <a shape="rect" href="http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient" style="color:#4453ea;" rel="nofollow" target="_blank">www.avg.com</a>
                          </td></tr></tbody></table>
                    <a shape="rect" href="#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2" rel="nofollow" target="_blank"> </a></div>
                </div>
              </div>
            </div>
            <div class="ydpb6a32dd9yiv8583323751ydp50d5cd69yqt3873327189" id="ydpb6a32dd9yiv8583323751ydp50d5cd69yqtfd87137">----<br clear="none">
              Election-Methods mailing list - see <a shape="rect" href="https://electorama.com/em " rel="nofollow" target="_blank">https://electorama.com/em
              </a>for list info<br clear="none">
            </div>
          </div>
        </div>
      </div>
    
  </div></div></div></div>
            </div>
        </div></body></html>