<html>
  <head>
    <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <div class="moz-cite-prefix">On 10/20/2016 6:35 AM, Michael Ossipoff
      wrote:<br>
      <blockquote type="cite">An example is needed.</blockquote>
      <br>
      C: Not really. Logically it's impossible to have both any
      Push-over incentive and FBC.<br>
      <br>
      Whenever you're determining the winner by who is pairwise
      preferred out of the winner of method A and the winner of method
      B,<br>
      there will always be situations where you do better by not
      equal-top voting your sincere favourite F so as to keep F out of
      the final<br>
      where F would lose to your worst W.<br>
      <br>
      Chris  Benham<br>
      <br>
      <br>
    </div>
    <blockquote
cite="mid:CAOKDY5AJ_V2hZXXsoZZ+o1HXZFfv0YbS_DMQwNkuxxUHVZEW9w@mail.gmail.com"
      type="cite">
      <p dir="ltr">An example is needed.</p>
      <div class="gmail_quote">On Oct 18, 2016 9:37 PM, "C.Benham" <<a
          moz-do-not-send="true" href="mailto:cbenham@adam.com.au">cbenham@adam.com.au</a>>
        wrote:<br type="attribution">
        <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_3547428967173254496moz-cite-prefix">On
              10/19/2016 8:10 AM, Michael Ossipoff wrote:<br>
              <br>
              <blockquote type="cite">It should be MMPO, rather than
                Smith//MMPO, for one finalist-choosing method, and
                Approval, Inferred-Approval, or Score for the other,
                because MMPO, Approval, & Score meet FBC.</blockquote>
              <br>
              FBC won't survive any Push-over incentive  (as I'm sure
              Kevin Venzke would confirm).<br>
              <br>
              Chris Benham<br>
              <br>
            </div>
            <blockquote type="cite">
              <p dir="ltr">It should be MMPO, rather than Smith//MMPO,
                for one finalist-choosing method, and Approval,
                Inferred-Approval, or Score for the other, because MMPO,
                Approval, & Score meet FBC.</p>
              <p dir="ltr">If Plain MMPO were replaced by anything else,
                Weak CD would be lost.</p>
              <p dir="ltr">If, for the other finalist-choosing method,
                Approval, Inferred-Approval or Score were replaced by
                MAM or Beatpath, then both finalist-choosing methods
                would share the same strategic vulnerabilities.</p>
              <p dir="ltr">Michael Ossipoff</p>
              <div class="gmail_quote">On Oct 18, 2016 1:42 PM, "Forest
                Simmons" <<a moz-do-not-send="true"
                  href="mailto:fsimmons@pcc.edu" target="_blank">fsimmons@pcc.edu</a>>
                wrote:<br type="attribution">
                <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>I appreciate all of the great
                                      insights from Kristofer, Chris
                                      Benham, and Michael Ossipoff.<br>
                                      <br>
                                    </div>
                                    Especially thanks to Kristofer for
                                    being a good sport about my
                                    forwarding an email with his private
                                    earlier input included.  It was too
                                    late when I realized I hadn't
                                    deleted that part.<br>
                                    <br>
                                  </div>
                                  <div>Intuitively, I think Chris is
                                    right that Pushover is the biggest
                                    potential problem.  But I don't see
                                    an obvious example.<br>
                                  </div>
                                  <div><br>
                                  </div>
                                  Michael is right that we need to
                                  consider other possibilities for the
                                  two base methods for picking the
                                  finalists.<br>
                                  <br>
                                </div>
                                I like MMPO or Smith//MMPO as one of
                                them since MMPO is one method that
                                doesn't just reduce to Approval when all
                                candidates are ranked or rated at the
                                extremes.  I think that the other method
                                should be one that does reduce to
                                Approval at the extremes, like River,
                                MAM/RankedPairs, or
                                Beatpath/Tideman/Schulz.  It could be a
                                Bucklin variant like MJ, Andy Jennings's
                                Chiastic Approval, or Jameson's MAS. <br>
                                <br>
                                 Like Michael I think that Range itself
                                gives too much incentive to vote at the
                                extremes on the strategic ballots. 
                                Better to use Approval or an approval
                                variant so that the strategic ratings
                                are not unduly compressed for the other
                                base method. <br>
                                <br>
                              </div>
                              I like Kristofer's insights about the
                              subtle differences between the proposed
                              "manual" version in contradistinction to a
                              DSV version that automates strategy for
                              the two methods based on the first set of
                              (perhaps somewhat pre-strategized)
                              ratings.<br>
                              <br>
                            </div>
                            In particular he pointed out how certain
                            procedural rules can externalize the
                            paradoxes of voting.  To a certain extent
                            Approval avoids bad properties by
                            externalizing them.  The cost is the
                            "burden" of the voter deciding whom to
                            approve.  As Ron LeGrand has so amply
                            demonstrated, any time you try to automate
                            approval strategy in a semi-optimal way, you
                            end up with a non-monotone method.  By the
                            same token IRV can be thought of as a
                            rudimentary DSV approach to plurality
                            voting, so it should be no surprise that
                            IRV/STV is non-monotone.<br>
                            <br>
                          </div>
                          A better example, closer to the Kristofer's,
                          idea is Asset Voting.  It externalizes
                          everything, which makes it impossible to
                          contradict any nice ballot based property. 
                          Because of this there is an extreme resulting
                          strategic burden, but in this case that burden
                          is placed squarely onto the shoulders of the
                          candidates, not the voters. Presumably the
                          candidates are up to that kind of burden since
                          they are, after all, politicians (in our
                          contemplated public applications).<br>
                          <br>
                        </div>
                        But this brings up another intriguing idea.  Let
                        one of the two base methods be Asset Voting, so
                        that the sincere ballots decide between (say)
                        the MMPO winner and the Asset Voting winner.<br>
                        <br>
                      </div>
                      Thanks Again,<br>
                      <br>
                    </div>
                    Forest<br>
                    <div class="gmail_extra"><br>
                      <div class="gmail_quote">On Tue, Oct 18, 2016 at
                        12:32 PM, Michael Ossipoff <span dir="ltr"><<a
                            moz-do-not-send="true"
                            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">
                          <p dir="ltr">If course the balloting for
                            choosing between the 2 finalists need only
                            be rankings, to show preferences between the
                            2 finalists, whoever they turn out to be.</p>
                          <p dir="ltr">Some variations occurred to me.
                            I'm not saying that any of them would be
                            better. I just wanted to mention them,
                            without any implication that they haven't
                            already occurred to everyone.</p>
                          <p dir="ltr">Both of the following
                            possibilities have disadvantages, in
                            comparison to the initial proposal:</p>
                          <p dir="ltr">1. What if, for the initial 2
                            counts, it were a Score-count, in addition
                            to the MMPO count.</p>
                          <p dir="ltr">One argument against that
                            variation is that a voter's inferred
                            approvals are likely to be more optimal for
                            hir than the Score ratings on which they're
                            based.</p>
                          <p dir="ltr">2. For the 2 initial counts, what
                            if the MMPO count used a separate ranking,
                            & the Approval count used a separate set
                            of Approval-marks?</p>
                          <p dir="ltr">Would that make it easier for
                            Chris's pushover strategist?</p>
                          <p dir="ltr">What other positive &
                            negative results?</p>
                          <p dir="ltr">One possible disadvantage that
                            occurs to me is that overcompromising voters
                            might approve lower than than necessary, if
                            the approval were explicitly voted.  ...in
                            comparison to their ratings-which tend to
                            soften voting errors.</p>
                          <p dir="ltr">So far, it appears that the
                            initial proposal is probably the best one.</p>
                          <span
                            class="m_3547428967173254496m_2236720442496602401HOEnZb"><font
                              color="#888888">
                              <p dir="ltr">Michael Ossipoff</p>
                            </font></span>
                          <div
                            class="m_3547428967173254496m_2236720442496602401HOEnZb">
                            <div
                              class="m_3547428967173254496m_2236720442496602401h5">
                              <div class="gmail_quote">On Oct 17, 2016
                                1:49 PM, "Forest Simmons" <<a
                                  moz-do-not-send="true"
                                  href="mailto:fsimmons@pcc.edu"
                                  target="_blank">fsimmons@pcc.edu</a>>
                                wrote:<br type="attribution">
                                <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>Kristofer,<br>
                                            <br>
                                            Perhaps the way out is to
                                            invite two ballots from each
                                            voter. The first set of
                                            ballots is used to narrow
                                            down to two alternatives. 
                                            It is expected that these
                                            ballots will be voted with
                                            all possible manipulative
                                            strategy ... chicken
                                            defection, pushover, burial,
                                            etc.<br>
                                            <br>
                                          </div>
                                          The second set is used only to
                                          decide between the two
                                          alternatives served up by the
                                          first set.<br>
                                          <br>
                                        </div>
                                        A voter who doesn't like
                                        strategic burden need not
                                        contribute to the first set, or
                                        could submit the same ballot to
                                        both sets.<br>
                                        <br>
                                      </div>
                                      <div>If both ballots were Olympic
                                        Score style, with scores ranging
                                        from blank (=0) to 10, there
                                        would be enough resolution for
                                        all practical purposes. 
                                        Approval voters could simply
                                        specify their approvals with 10
                                        and leave the other candidates'
                                        scores blank.<br>
                                        <br>
                                      </div>
                                      <div>There should be no
                                        consistency requirement between
                                        the two ballots.  They should be
                                        put in separate boxes and
                                        counted separately.  Only that
                                        policy can guarantee the
                                        sincerity of the ballots in the
                                        second set.<br>
                                        <br>
                                      </div>
                                      <div>In this regard it is
                                        important to realize that
                                        optimal perfect information
                                        approval strategy may require
                                        you to approve out of order,
                                        i.e. approve X and not Y even if
                                        you sincerely rate Y higher than
                                        X.  [We're talking about optimal
                                        in the sense of maximizing your
                                        expectation, meaning the
                                        expectation of your sincere
                                        ratings ballot, (your
                                        contribution to the second
                                        set).] <br>
                                        <br>
                                      </div>
                                      <div>Nobody expects sincerity on
                                        the first set of ballots.  If
                                        some of them are sincere, no
                                        harm done, as long as the
                                        methods for choosing the two
                                        finalists are reasonable.<br>
                                        <br>
                                      </div>
                                      <div>On the other hand, no
                                        rational voter would vote
                                        insincerely on hir contribution
                                        to the second set.  The social
                                        scientist has a near perfect
                                        window into the sincere
                                        preferences of the voters.<br>
                                        <br>
                                      </div>
                                      <div>Suppose the respective
                                        finalists are chosen by IRV and
                                        Implicit Approval, respectively,
                                        applied to the first set of
                                        ballots.  People's eyes would be
                                        opened when they saw how often
                                        the Approval Winner was
                                        sincerely preferred over the IRV
                                        winner.<br>
                                        <br>
                                      </div>
                                      <div>Currently my first choice of
                                        methods for choosing the
                                        respective finalists would be
                                        MMPO for one of them and
                                        Approval for the other, with the
                                        approval cutoff at midrange (so
                                        scores of six through ten
                                        represent approval).<br>
                                        <br>
                                      </div>
                                      <div>Consider the strategical
                                        ballot set profile conforming to<br>
                                        <br>
                                      </div>
                                      <div>40  C<br>
                                      </div>
                                      <div>32  A>B<br>
                                      </div>
                                      <div>28  B<br>
                                        <br>
                                      </div>
                                      <div>The MMPO finalist would be A,
                                        and the likely Approval finalist
                                        would be B, unless too many B
                                        ratings were below midrange.<br>
                                        <br>
                                      </div>
                                      <div>If the sincere ballots were<br>
                                        <br>
                                      </div>
                                      <div>40 C<br>
                                      </div>
                                      <div>32 A>B<br>
                                      </div>
                                      <div>28 B>A<br>
                                        <br>
                                      </div>
                                      <div>then the runoff winner
                                        determined by the second set of
                                        ballots would be A, the CWs. 
                                        The chicken defection was to no
                                        avail.  Note that even though
                                        this violates Plurality on the
                                        first set of ballots, it does
                                        not on the sincere set.<br>
                                        <br>
                                      </div>
                                      <div>On the other hand, if the
                                        sincere set conformed to<br>
                                        <br>
                                      </div>
                                      <div>40 C>B<br>
                                      </div>
                                      <div>32 A>B<br>
                                      </div>
                                      <div>28 B>C<br>
                                        <br>
                                      </div>
                                      <div>then the runoff winner would
                                        be B, the CWs, and the C faction
                                        attempt to win by truncation of
                                        B would have no effect.  A
                                        burial of B by the C faction
                                        would be no more rewarding than
                                        their truncation of B.<br>
                                        <br>
                                      </div>
                                      <div>So this idea seems to take
                                        care of the tension between
                                        methods that are immune to
                                        burial and methods that are
                                        immune to chicken defection.<br>
                                        <br>
                                      </div>
                                      <div>Furthermore, the plurality
                                        problem of MMPO evaporates. 
                                        Even if all of the voters vote
                                        approval style in either or both
                                        sets of ballots, the Plurality
                                        problem will automatically
                                        evaporate; on approval style
                                        ballots the Approval winner
                                        pairwise beats all other
                                        candidates, including the MMPO
                                        candidate (if different from the
                                        approval winner).<br>
                                        <br>
                                      </div>
                                      <div>What do you think?<br>
                                        <br>
                                      </div>
                                      <div>Forest<br>
                                      </div>
                                      <div><br>
                                        <br>
                                      </div>
                                      <div><br>
                                      </div>
                                      <br>
                                    </div>
                                    <div class="gmail_extra"><br>
                                      <div class="gmail_quote">On Sun,
                                        Oct 16, 2016 at 1:30 AM,
                                        Kristofer Munsterhjelm <span
                                          dir="ltr"><<a
                                            moz-do-not-send="true"
                                            href="mailto:km_elmet@t-online.de"
                                            target="_blank">km_elmet@t-online.de</a>></span>
                                        wrote:<br>
                                        <blockquote class="gmail_quote"
                                          style="margin:0 0 0
                                          .8ex;border-left:1px #ccc
                                          solid;padding-left:1ex"><span>On
                                            10/15/2016 11:56 PM, Forest
                                            Simmons wrote:<br>
                                            > Thanks, Kristofer; it
                                            seems to be a folk theorem
                                            waiting for formalization.<br>
                                            ><br>
                                            > That reminds me that
                                            someone once pointed out
                                            that almost all of the<br>
                                            > methods favored by EM
                                            list enthusiasts reduce to
                                            Approval when only top<br>
                                            > and bottom votes are
                                            used, in particular when
                                            Condorcet methods allow<br>
                                            > equal top and multiple
                                            truncation votes they fall
                                            into this category<br>
                                            > because the Approval
                                            Winner is the pairwise
                                            winner for approval style<br>
                                            > ballots.<br>
                                            ><br>
                                            > Everything else
                                            (besides approval strategy)
                                            that we do seems to be an<br>
                                            > effort to lift the
                                            strategical burden from the
                                            voter.  We would like to<br>
                                            > remove that burden in
                                            all cases, but at least in
                                            the zero info case.<br>
                                            > Yet that simple goal is
                                            somewhat elusive as well.<br>
                                            <br>
                                          </span>Suppose we have a proof
                                          for such a theorem. Then you
                                          could have a<br>
                                          gradient argument going like
                                          this:<br>
                                          <br>
                                          - If you're never harmed by
                                          ranking Approval style, then
                                          you should do so.<br>
                                          - But figuring out the correct
                                          threshold to use is tough
                                          (strategic burden)<br>
                                          - So you may err, which leads
                                          to a problem. And even if you
                                          don't, if<br>
                                          the voters feel they have to
                                          burden their minds, that's a
                                          bad thing.<br>
                                          <br>
                                          Here, traditional game theory
                                          would probably pick some kind
                                          of mixed<br>
                                          strategy, where you
                                          "exaggerate" (Approval-ize)
                                          only to the extent that<br>
                                          you benefit even when taking
                                          your errors into account. But
                                          such an<br>
                                          equilibrium is unrealistic
                                          (we'd have to find out why,
                                          but probably<br>
                                          because it would in the worst
                                          case require everybody to know
                                          about<br>
                                          everybody else's level of
                                          bounded rationality).<br>
                                          <br>
                                          And if the erring causes
                                          sufficiently bad results,
                                          we're left with two<br>
                                          possibilities:<br>
                                          <br>
                                          - Either suppose that the
                                          method is sufficiently robust
                                          that most voters<br>
                                          won't use Approval strategy
                                          (e.g. the pro-MJ argument that
                                          Approval<br>
                                          strategy only is a benefit if
                                          enough people use it, so most
                                          people<br>
                                          won't, so we'll have a
                                          correlated equilibrium of
                                          sorts)<br>
                                          <br>
                                          - That any admissible method
                                          must have a "bump in the road"
                                          on the way<br>
                                          from a honest vote to an
                                          Approval vote, where moving
                                          closer to<br>
                                          Approval-style harms the
                                          voter. Then a game-theoretical
                                          voter only votes<br>
                                          Approval style if he can
                                          coordinate with enough other
                                          voters to pass the<br>
                                          bump, which again is
                                          unrealistic.<br>
                                          <br>
                                          But solution #2 will probably
                                          destroy quite a few nice
                                          properties (like<br>
                                          monotonicity + FBC; if the
                                          proof is by contradiction,
                                          then we'd know<br>
                                          some property combinations
                                          we'd have to violate). So we
                                          can't have it all.<br>
                                        </blockquote>
                                      </div>
                                      <br>
                                    </div>
                                  </div>
                                </blockquote>
                              </div>
                            </div>
                          </div>
                        </blockquote>
                      </div>
                      <br>
                    </div>
                  </div>
                </blockquote>
              </div>
              <br>
              <fieldset
                class="m_3547428967173254496mimeAttachmentHeader"></fieldset>
              <br>
              <pre>----
Election-Methods mailing list - see <a moz-do-not-send="true" class="m_3547428967173254496moz-txt-link-freetext" href="http://electorama.com/em" target="_blank">http://electorama.com/em</a> for list info
</pre>
              <br>
              <fieldset
                class="m_3547428967173254496mimeAttachmentHeader"></fieldset>
              <br>
              <br>
            </blockquote>
          </div>
          <br>
        </blockquote>
      </div>
      <p class="" avgcert""="" color="#000000" align="left"><br>
      </p>
    </blockquote>
    <p><br>
    </p>
  </body>
</html>