<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote"><br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div>Of the methods that combine Approval with pairwise-count, maybe PAV is the one that is most favorable to Approval, while still letting people vote rankings that count and letting pairwise-count have a role.<br><br></div><div>Does that make it the best of that class?<span class="HOEnZb"><font color="#888888"><br></font></span></div><span class="HOEnZb"><font color="#888888"><div><br></div>Michael Ossipoff<br></font></span></div><div class="HOEnZb"><div class="h5"><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Nov 17, 2016 at 1:54 PM, Michael Ossipoff <span dir="ltr"><<a href="mailto:email9648742@gmail.com" target="_blank">email9648742@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div>But wouldn't Smith//Approval, with approval cutoffs in the rankings, share MDDTR's burial-vullnerability?<br><br></div>...with, additionally, vulnerability to truncation, which MDDTR _doesn't_ have?<br><br></div>And Smith//Approval trades MDDTR's FBC for Smith, which I consider an unfavorable trade.<br><br></div>I now again consider Benham & Woodall to be the best proposals for an electorate that wants &/or needs rankings...because of the IRV mitigations that i mentioned in my previous posting, and because of the context of the always high price of CD, as exemplified by MDDTR.<br><br></div>It could be debatable whether IRV's CWs elimination problem is worse than MDDTR's burial vulnerability problem, but, because of IRV's mitigations, I tend to feel that IRV is better, _for people who want or need routine ranking_.<br></div>Myself, I personally prefer MDDTR, with middle-ranking usually avoided, but I'm talking about what's best for the electorate as a whole.<br><br></div>In polls with balloting by both Approval & Score, I've observed overcompromisers doing better with Score than with Approval. They (and rival parties) would probably do even better with rankings.<br><br></div>Suggested merit-order for electorates that want &/or need rankings:<br><br></div>1.Benham<br></div>2. woodall<br></div>3. IRV<br></div>4. MDDTR<br></div>5. Methods combining Approval with pairwise-count<br></div>6. ER Bucklin/MJ<br></div>7. Score<br></div>8. Approval<br><br></div>I still personally prefer Approval, or maybe MDDTR, when used as Approval with 2nd-ranking as a defection-deterrent. But this suggested merit-order is for electorates who want &/or need rankings.<br><br></div>On pure merit, Woodall seems a bit better than Benham, because it's more particular which Smith-set member it elects. But Benham is much briefer to define, not needing to define or mention the Smith-set, and that makes a big difference in proposability, which outweighs the small merit-difference.<br><br></div>I don't know the merit-order among the methods that combine Approval with pairwise-count.<br><br>Maybe the ones that start with Approval are better than the ones that start with pairwise-count. <br><br></div>Among those, Brams' PAV is vulnerable to truncation, where MAMPO isn't.<br><br></div>But maybe that's a good thing, if the majority CWs among the majority-approved candidates isn't in your strong top-set.<br><br></div>A proposal that the Greens upgrade from IRV to Benham seems reasonably proposable.<br><br></div>Though Benham & Woodall are vulnerable to burial & truncation, the worst that can result from those strategies is the same as what IRV would have done anyway.<span class="m_-930797630882006229HOEnZb"><font color="#888888"><br><br></font></span><div><div><span class="m_-930797630882006229HOEnZb"><font color="#888888">Michael Ossipoff</font></span><div><div class="m_-930797630882006229h5"><br><div><div><div><br><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><div><br><br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Nov 17, 2016 at 9:05 AM, C.Benham <span dir="ltr"><<a href="mailto:cbenham@adam.com.au" target="_blank">cbenham@adam.com.au</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
  
    
  
  <div bgcolor="#FFFFFF" text="#000000">
    <div class="m_-930797630882006229m_-5128077215961204172m_-9006749870759545372moz-cite-prefix"><span>On 11/17/2016 9:00 AM, Forest Simmons
      wrote:<br>
      <br>
      <blockquote type="cite">
        <div>
          <div>Here's a simple method that is essentially
            Smith//Approval without having to mention the Smith set:<br>
            <br>
          </div>
          List the candidates in order of approval, highest to lowest,
          top to bottom.  While any candidate pairwise beats an adjacent
          candidate higher in the list, switch places of the two lowest
          out of order adjacent members.<br>
          <br>
        </div>
        When there remains no out of order adjacent pair, elect the
        candidate at the top of the list.</blockquote>
      <br></span>
      Forest,<br>
      <br>
      I like Smith//Approval  and your usually equivalent Max Covered
      Approval method.  <br>
      <br>
      But in this version, the stipulation that we  "switch places of
      the two <i>lowest</i> out of order adjacent members" could look a
      bit arbitrary and less smooth<br>
      than Margins-Sorted Approval.<br>
      <br>
      BTW, it seems to me that both this and  Smith//Approval  can
      handle the  Chicken Dilemma situation quite well if we use ranked
      ballots with  approval cut-offs.<br>
      <br>
      (Or ratings ballots with many  slots that register approval and as
      many (or maybe as few as only 2) that register unapproval.)<br>
      <br>
      <a class="m_-930797630882006229m_-5128077215961204172m_-9006749870759545372moz-txt-link-freetext" href="http://wiki.electorama.com/wiki/Approval_Sorted_Margins" target="_blank">http://wiki.electorama.com/wik<wbr>i/Approval_Sorted_Margins</a><br>
      <br>
      <a class="m_-930797630882006229m_-5128077215961204172m_-9006749870759545372moz-txt-link-freetext" href="http://wiki.electorama.com/wiki/Approval_Cutoff" target="_blank">http://wiki.electorama.com/wik<wbr>i/Approval_Cutoff</a><br>
      <br>
      And here's another smooth Condorcet method that should do as well:<br>
      <br>
      *Voters  score the candidates on some scale that allows large and
      varied gaps between the candidates: say 0-100.<br>
      Elect the CW if there is one.<br>
      <br>
      Otherwise compress the 1-point gaps (if any) on all ballots into
      zero-point gaps (so that those ballots abandon their original
      pairwise preference for any <br>
      X originally scored only one point more than any Y).<br>
      <br>
      Based on the thus modified ballot information, elect the CW if
      there is one.<br>
      <br>
      Otherwise compress the 2-point gaps (if any) on all ballots into
      zero-point gaps (so that those ballots abandon their original
      pairwise preference for any <br>
      X originally scored two points more than any Y).<br>
      <br>
      Based on the thus modified ballot information, elect the CW if
      there is one.<br>
      <br>
      And so on, as gradually as possible compressing larger and larger
      gaps until we have a pairwise beats-all winner.*<br>
      <br>
      Chris Benham<div><div class="m_-930797630882006229m_-5128077215961204172h5"><br>
      <br>
      <br>
      On 11/17/2016 9:00 AM, Forest Simmons wrote:<br>
    </div></div></div><div><div class="m_-930797630882006229m_-5128077215961204172h5">
    <blockquote type="cite">
      <div dir="ltr">
        <div>
          <div>
            <div>
              <div>Here's a simple method that is essentially
                Smith//Approval without having to mention the Smith set:<br>
                <br>
              </div>
              List the candidates in order of approval, highest to
              lowest, top to bottom.  While any candidate pairwise beats
              an adjacent candidate higher in the list, switch places of
              the two lowest out of order adjacent members.<br>
              <br>
            </div>
            When there remains no out of order adjacent pair, elect the
            candidate at the top of the list.<br>
            <br>
          </div>
          Note that the winner will automatically be a member of the top
          cycle, and if it is a cycle of three, it will be the most
          approved member of the cycle.<br>
          <br>
        </div>
        Also notice that it yields an unambiguous social order, and that
        there can be no second place complaint.<br>
        <br>
        <br>
        <br>
        <br>
      </div>
    </blockquote>
    <blockquote type="cite">
      <div class="gmail_extra"><br>
        <div class="gmail_quote">On Tue, Nov 15, 2016 at 5:19 PM,
          Michael Ossipoff <span dir="ltr"><<a href="mailto:email9648742@gmail.com" target="_blank">email9648742@gmail.com</a>></span>
          wrote:<br>
          <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <div dir="ltr">
              <div>
                <div>
                  <div>
                    <div>
                      <div>
                        <div>
                          <div>
                            <div>
                              <div>
                                <div>
                                  <div>
                                    <div>
                                      <div>
                                        <div>
                                          <div>
                                            <div>
                                              <div>
                                                <div>
                                                  <div>When I started my
                                                    current EM
                                                    participation, I was
                                                    saying that 3-Slot
                                                    ICT was my favorite
                                                    method.<br>
                                                    <br>
                                                  </div>
                                                  That doesn't conflict
                                                  with saying that I
                                                  consider Approval the
                                                  best, because I regard
                                                  3-Slot ICT, or
                                                  unlimited-rankings ICT
                                                  (when used
                                                  approval-like) as an
                                                  Approval version
                                                  without
                                                  chicken-dilemma.<br>
                                                  <br>
                                                </div>
                                                Later I realized that
                                                MDDTR is better than
                                                ICT, because it gives
                                                better protection to
                                                middle candidates.<br>
                                                <br>
                                              </div>
                                              <div>I measure that
                                                protection by how well
                                                they'd be protected if
                                                they were CWs.  ...what
                                                it would take to protect
                                                their win, and how well
                                                it's protected.<br>
                                              </div>
                                              <div><br>
                                              </div>
                                              I define "middle
                                              candidates" as candidates
                                              you rank or rate below top
                                              and above bottom.<br>
                                              <br>
                                            </div>
                                            ICT gives no protection to
                                            middle candidates, against
                                            burial, or even against
                                            innocent, non-strategic
                                            truncation--the two things
                                            that threaten a CWs in
                                            pairwise-c0unt methods.<br>
                                            <br>
                                          </div>
                                          MDDTR gives full
                                          truncation-proofness to middle
                                          candidates, but (contrary to
                                          what I earlier believed), its
                                          protection of middle
                                          candidates against burial can
                                          only be called "shabby".<br>
                                        </div>
                                        <br>
                                      </div>
                                      By the way, I no longer think that
                                      ICT or MDDTR needs to be 3-slot.
                                      3-Slot would be fine with me,
                                      because I believe that ICT or
                                      MDDTR should be used as Approval,
                                      and that middle rating or ranking
                                      should only be used when seriously
                                      needed to deter chicken-dilemma
                                      defection. When middle is used in
                                      an unlimited-ranking MDDTR or ICT,
                                      it should probably consist of
                                      2nd-place ranking, if you want to
                                      give the demoted candidates the
                                      best protection.   ...but maybe
                                      you'd rather rank them with
                                      respect to eachother, at different
                                      middle levels, as I probably
                                      sometimes would.<br>
                                      <br>
                                    </div>
                                    But, as I've been saying, activists
                                    & organizations seem to like
                                    rankings, and some
                                    people--overcompromisers & rival
                                    parties--might very well need
                                    rankings to soften their voting
                                    errors.<br>
                                    <br>
                                  </div>
                                  And it seems to me that there's no
                                  particular reason not to rank, in
                                  order of preference, your middle
                                  candidates, if some of them are better
                                  than others, or if the voters of some
                                  of them are less trustworthy than
                                  others.<br>
                                  <br>
                                </div>
                                So, that's if you want CD, in addition
                                to FBC, and good protection for middle
                                candidates<br>
                                <br>
                              </div>
                              Even if you're using the method as
                              Approval, you still want your demoted
                              candidate(s) to be well protected. Just
                              because you don't trust hir voters doesn't
                              mean you want to throw her to the hounds
                              and thereby lower Pt, the probability of
                              electing from your strong top-set.<br>
                              <br>
                            </div>
                            Anyway, so far, this is all referring to CD
                            methods.<br>
                            <br>
                          </div>
                          Of those, I like MDDTR best. As a rank method,
                          it (as i said) gives only shabby
                          burial-protection to a middle candidate. But
                          evidently (please tell me it isn't so) you
                          can't have FBC, CD, and good protection of
                          middle candidates.<br>
                          <br>
                        </div>
                        I consider CD more important to how well
                        protected middle candidates are. Yes, FBC + CD
                        give poor protection to middle candidates, and
                        that lessens the value of their CD. But non-CD
                        methods don't have CD at all, and that's worse.<br>
                        <br>
                      </div>
                      So I prefer MDDTR to methods that give better
                      protection to middle candidates, but don't have
                      CD.<br>
                      <br>
                    </div>
                    So, where I used to say that my favorite method is
                    3-Slot ICT, now I say that my favorite method is
                    MDDTR. Preferably with unlimited rankings. (Though
                    one could use only the 1st, 2nd, & bottom
                    positions if one chose to).   ...regardable as a
                    chicken-dilemma-free version of Approval.<br>
                    <br>
                    ------------------------------<wbr>------------------------------<wbr>------<br>
                    <br>
                  </div>
                  <div>Non-CD methods with better "middle-strategy" than
                    CD methods:<br>
                    <br>
                  </div>
                  But, in an election, I'm just one voter, and so, how
                  well-suited the method is to me is less important, and
                  won't affect the outcome as much, in comparison to how
                  well-suited the method is to lots of progressives. <br>
                  <br>
                </div>
                So, what if most progressives would rather have a method
                that's really good as a rank method, a method that has
                good "middle strategy" (strategy for protecting a middle
                candidate's win if s/he's CWs). <br>
                <br>
              </div>
              That would be important if you knew that all or nearly
              all, or even most of them were going to use the method
              purely as a rank method.<br>
              <div>
                <div>
                  <div><br>
                  </div>
                  <div>Bucklin is the traditional FBC rankings-method.<br>
                    <br>
                  </div>
                  <div>I distinguish 2 kinds of middle strategy merit:<br>
                    <br>
                  </div>
                  <div>1. How well the method protects top-ranked
                    candidates against middle-ranked candidates. I call
                    that "Middle1"<br>
                    <br>
                  </div>
                  <div>2. How well the lmethod  protects a middle-ranked
                    candidate against any candidate you rank lower than
                    hir. I call that "Middle2".<br>
                    <br>
                  </div>
                  <div>So, how to get the best middle strategy, with the
                    main goal still being keeping a good probability,
                    Pt, of electing from your strong top-set?<br>
                    <br>
                  </div>
                  <div>MDDTR's middle1 seems better than that of
                    Bucklin. In MDDTR, you're voting to contribute to a
                    majority for your top against your middle. In
                    Bucklin, you can protect top against middle by
                    skipping some rating-levels above the middle
                    candidates. In that way, you can give the top
                    candidates time to receive the coalescing
                    lower-choice votes that they'll get from the
                    preferrers of other candidates, before giving
                    anything to the middle candidates.<br>
                    <br>
                  </div>
                  <div>That's a bit more work than just ranking in order
                    of preference. It requires you to judge where, and
                    how far down in rankings, your top candidates are
                    going to receive lower-choice votes from.<br>
                    <br>
                  </div>
                  <div>So I suggest that MDDTR does better at Middle1
                    than Bucklin does.<br>
                    <br>
                  </div>
                  <div>But Bucklin does better at Middle2.<br>
                    <br>
                  </div>
                  <div>In Bucklin, the CWs's win is protected by the
                    people who pretty-much agree with you, the people of
                    your wing, merely not ranking down too far. <br>
                    <br>
                  </div>
                  <div>MDDTR needs that too, but it isn't enough to give
                    MDDTR more than shabby protection.<br>
                  </div>
                  <div>
                    <div>
                      <div>
                        <div>
                          <div>
                            <div>
                              <div>
                                <div>
                                  <div>
                                    <div><br>
                                      <div>
                                        <div>...And Bucklin's Middle1,
                                          though not as convenient or
                                          easy as that of MDDTR, isn't
                                          as questionable as MDDTR's
                                          Middle2. <br>
                                          <br>
                                        </div>
                                        <div>So, overall, I'd say that
                                          Bucklin's Middle Strategy is
                                          better than that of MDDTR. So,
                                          for people who want to use the
                                          method purely as a
                                          rank-method, Bucklin is better
                                          than MDDTR.<br>
                                          <br>
                                        </div>
                                        <div>Bucklin also has the
                                          advantage of use-precedent. 
                                          MDDT has the advantage of
                                          precinct-summability,but I
                                          don't consider that essential.
                                          <br>
                                          <br>
                                        </div>
                                        <div>For voters using the method
                                          purely as a rank method, I'd
                                          prefer Bucklin to MDDTR.<br>
                                          <br>
                                        </div>
                                        <div>Chicken dilemma won't
                                          happen all the time, probably
                                          won't happen often. But
                                          middle-protection will always
                                          matter to people using it as a
                                          rank method.<br>
                                          <br>
                                        </div>
                                        <div>But it seems to me that,
                                          once we give up CD (for voters
                                          who need good middle
                                          strategly, because of their
                                          rank voting), then it might be
                                          possible to do better than
                                          Bucklin.<br>
                                          <br>
                                        </div>
                                        <div>It seems to me that methods
                                          that use both Approval and
                                          pairwise-count can do better
                                          than Bucklin, at middle
                                          protection.<br>
                                          <br>
                                        </div>
                                        <div>A lot of methods of that
                                          kind have been proposed, and
                                          I've ignored all of them
                                          because they don't meet CD.
                                          But, as mentioned above, for
                                          some electorates, middle
                                          strategy could be more
                                          important.<br>
                                          <br>
                                        </div>
                                        <div>It seems to me that MDDA
                                          (also evidently named MPOA)
                                          and Smith//Approval are two
                                          methods that might be better
                                          than Bucklin at middle
                                          protection..<br>
                                          <br>
                                        </div>
                                        <div>Using Approval as the
                                          cycle-solution is a very
                                          powerful idea (if you're
                                          willing to give up CD, for an
                                          electorate's needs). But most
                                          of you already knew that,
                                          before I paid attention to it
                                          (...because I was only looking
                                          at CD methods)..<br>
                                          <br>
                                        </div>
                                        <div>MDDA's &
                                          Smith//Approval's burial
                                          vulnerability doesn't matter
                                          much, when the Approval winner
                                          wins the cycle. In fact,
                                          Smith//Approval's
                                          truncation-vulnerability could
                                          even be regarded as an
                                          advantage, for when your
                                          strong top-set doesn't include
                                          the CWs.<br>
                                          <br>
                                        </div>
                                        <div>MDDA & Smith//Approval
                                          look better to me than
                                          Bucklin.<br>
                                          <br>
                                        </div>
                                        <div>Simpler Middle1. <br>
                                          <br>
                                        </div>
                                        <div>Precinct-Summability is an
                                          added bonus.<br>
                                          <br>
                                        </div>
                                        <div>MDDA seems to have a
                                          briefer definition than either
                                          Bucklin or Smith//Approval,
                                          and brief definition can be
                                          decisive.<br>
                                          <br>
                                        </div>
                                        <div>I know of Bucklin being
                                          rejected when MDDTR was
                                          accepted. MDDA would almost
                                          surely have been accepted too.<br>
                                          <br>
                                        </div>
                                        <div>I don't  think
                                          Smith//Approval would go over
                                          well, with its need to define
                                          the Smith set, which greatly
                                          lengthens the definition.<br>
                                          <br>
                                        </div>
                                        <div>For an electorate that need
                                          good Middle1 & Middle2
                                          more than CD, MDDA seems the
                                          winner so far.<br>
                                          <br>
                                        </div>
                                        <div>Smith//Approvsl of course
                                          meets Smith.  ...which of
                                          course means that it fails
                                          FBC. But does it need FBC?<br>
                                          <br>
                                        </div>
                                        <div>It could be argued (but I
                                          don't know if it's true) that
                                          Smith//Approval doesn't need
                                          FBC, because, though you don't
                                          have an efffective Approval
                                          vote at the top, you still can
                                          vote Approval, with the
                                          approval-cutoff, or by only
                                          ranking your strong top-set. <br>
                                          <br>
                                        </div>
                                        <div>So, though Compromise could
                                          become pair-beaten by Favorite
                                          because you raise Favorite to
                                          top with Compromise, resulting
                                          in a cycle instead of a CWv
                                          win for Favorite, the cycle
                                          will be judged by approvals,
                                          and you're approved only your
                                          strong top-set.<br>
                                        </div>
                                        <div><br>
                                          Of course, just because
                                          Favorite was almost the CWv
                                          doesn't necessarily mean that
                                          s/he'll win the Approval
                                          count. But are you any worse
                                          off than you'd have been with
                                          MDDA?<br>
                                          <br>
                                        </div>
                                        <div>Forest (but maybe others
                                          too) has proposed a number of
                                          methods that combine
                                          pairwise-count and Approval.
                                          Do any of those beat MDDA
                                          & Smith//Approval by the
                                          standards of protecting one's
                                          strong top-set, and Middle1
                                          & Middle2?<br>
                                          <br>
                                        </div>
                                        <div>in particular, do any of
                                          them do better than MDDA by
                                          those standards? Do any do as
                                          well as MDDA by those
                                          standards and have as brief a
                                          defintion, or nearly as brief
                                          a definition?<br>
                                          <br>
                                        </div>
                                        <div>In other words, are there
                                          methods that achieve those
                                          things better than MDDA &
                                          Smith//Approval, or achieve
                                          them better than MDDA and have
                                          as brief a definition?<br>
                                          <br>
                                        </div>
                                        <div>In fact, is there a method
                                          that meets FBC (or doesn't
                                          need it), meets CD, and does
                                          as well by Middle1 &
                                          Middle2 as MDDA,
                                          Smith//Approval or Bucklin?<span class="m_-930797630882006229m_-5128077215961204172m_-9006749870759545372HOEnZb"><font color="#888888"><br>
                                              <br>
                                            </font></span></div>
                                        <span class="m_-930797630882006229m_-5128077215961204172m_-9006749870759545372HOEnZb"><font color="#888888">
                                            <div>Michael Ossipoff<br>
                                            </div>
                                            <div><br>
                                              <br>
                                            </div>
                                            <div><br>
                                              <br>
                                              <br>
                                            </div>
                                            <div><br>
                                              <br>
                                              <br>
                                              <br>
                                              <br>
                                              <br>
                                              <br>
                                            </div>
                                            <div><br>
                                              <br>
                                            </div>
                                            <div><br>
                                              <br>
                                            </div>
                                          </font></span></div>
                                    </div>
                                  </div>
                                </div>
                              </div>
                            </div>
                          </div>
                        </div>
                      </div>
                    </div>
                  </div>
                </div>
              </div>
            </div>
          </blockquote>
        </div>
        <br>
      </div>
      <p color="#000000" align="left"><br>
      </p>
    </blockquote>
    <p><br>
    </p>
  </div></div></div>

</blockquote></div><br></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div>
</blockquote></div><br></div>
</div></div></blockquote></div><br></div></div>