<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p>Toby Pereira wrote:
      <blockquote type="cite">I'm not a fan of STAR, but I am still
        interested in seeing how it stands up to scrutiny given that it
        has a following. (Actually I'm not aware of how STAR fails
        monotonicity. I was under the impression that it passed.)
        <div dir="ltr" data-setdir="false"><br>
        </div>
      </blockquote>
    </p>
    <p>Toby,<br>
      <br>
      To give you a bit of a preview before I get around to cooking up
      all the examples, nothing with such obvious Push-over incentive
      can meet mono-raise (aka "monotonicty")<br>
      <br>
      Suppose  X beats Y in the final.   Now suppose on some ballots
      with Y above X, we raise X so it is now above Y.  That could
      reduce Y's score enough for it to be replaced in the final<br>
      by Z, a candidate that pairwise beats X.<br>
      <br>
      Voters who are mainly concerned to have their favourite X win and
      are fairly certain that X will reach the final will have a strong
      incentive to give X max points (5) and then also<br>
      give a 4 (or even a 5) to all those candidates that they think X
      can beat pairwise.<br>
    </p>
    <p>If enough voters use that strategy and it fails, both the
      finalists could be candidates with little sincere support.<br>
      <br>
      Chris Benham<br>
      <br>
    </p>
    <div class="moz-cite-prefix">On 15/08/2023 9:38 pm, Toby Pereira
      wrote:<br>
    </div>
    <blockquote type="cite"
      cite="mid:736465645.5313905.1692101284422@mail.yahoo.com">
      <meta http-equiv="content-type" content="text/html; charset=UTF-8">
      <div class="ydp79d2568dyahoo-style-wrap"
        style="font-family:Helvetica Neue, Helvetica, Arial,
        sans-serif;font-size:13px;">
        <div><br>
        </div>
        <div><br>
        </div>
      </div>
      <div id="ydpe5cc3871yahoo_quoted_2742889147"
        class="ydpe5cc3871yahoo_quoted">
        <div style="font-family:'Helvetica Neue', Helvetica, Arial,
          sans-serif;font-size:13px;color:#26282a;">
          <div> On Tuesday, 15 August 2023 at 06:28:51 BST, C.Benham
            <a class="moz-txt-link-rfc2396E" href="mailto:cbenham@adam.com.au"><cbenham@adam.com.au></a> wrote: </div>
          <div>
            <div id="ydpe5cc3871yiv4812049359">
              <div id="ydpe5cc3871yiv4812049359yqtfd24117"
                class="ydpe5cc3871yiv4812049359yqt5236108305">
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false">
                  <blockquote type="cite"><br>
                  </blockquote>
                  <blockquote type="cite">>>As for what the boxes
                    are about, they are about ensuring that voting
                    methods have sensible behaviour in certain
                    situations, so I wouldn't expect them to
                    >>necessarily negatively correlate with a
                    "good" result.
                    <div dir="ltr"><br clear="none">
                    </div>
                  </blockquote>
                  >Then you are not using the right boxes.<br
                    clear="none">
                  <br>
                  I think you probably misread my sentence. I would
                  *not* expect them to *negatively* correlate with a
                  good result. So I probably would expect passing
                  criteria to correlate positively with a good result.
                  Apologies for the convoluted sentence.<br clear="none">
                  <blockquote type="cite">
                    <div dir="ltr"><br clear="none">
                    </div>
                    >>It's what I just replied to Kristofer - <span><span
                        style="font-family:Helvetica Neue, Helvetica,
                        Arial, sans-serif;">the fact that there's no way
                        that you can consistently define society's
                        preference in a way that you can determine
                        whether >>society prefers A or B by
                        looking at the pairwise comparison.</span></span></blockquote>
                  <br clear="none">
                  >Then what do you "look at" ??</div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"><br>
                </div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false">There's different things
                  you can look at. But what I said is true, not just my
                  opinion, so it's a question for everyone. But you can
                  still look at pairwise comparisons and indeed use a
                  Condorcet method. But I'd want someone to come to this
                  by evaluating various methods and deciding that this
                  is the best compromise overall, rather than deciding
                  from the outset that this is the One True Way. I'm not
                  anti-Condorcet, if this is how it's come across.<br
                    clear="none">
                  <br clear="none">
                  <blockquote type="cite">
                    <div dir="ltr">>>I would counter this by
                      saying that it's based on this logical
                      contradiction.<br>
                    </div>
                  </blockquote>
                  <br clear="none">
                  >I'm still not seeing this "logical contradiction".</div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false"><br>
                </div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false">It's just the premise
                  that if A pairwise beats B then society must prefer A
                  to B. But it makes no sense to say that society
                  prefers A to B, B to C and C to A. I don't think this
                  is controversial. It's a centuries-old realisation.<br
                    clear="none">
                  <br clear="none">
                  <blockquote type="cite"><span><span
                        style="font-family:Helvetica Neue, Helvetica,
                        Arial, sans-serif;">>>I would also counter
                        it by saying that someone could equally say <span><span
                            style="color:rgb(38, 40,
                            42);font-family:Helvetica Neue, Helvetica,
                            Arial, sans-serif;">(perhaps have a greater
                            claim) </span></span>that a method cannot be
                        democratic if it fails participation</span></span></blockquote>
                  <br clear="none">
                  >I don't see that either. So which method that
                  meets Participation do you like?</div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false"><br>
                </div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false">I do actually like both
                  score and approval. I like their simplicity and the
                  fact that a complete numerical results list can be
                  published that's simple and easy to understand. And I
                  like the fact that there's no hidden weird paradoxes
                  in either method. Any criterion they fail is pretty
                  obvious.</div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false"><br>
                </div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false">My absolute favourite
                  method isn't suitable for many elections because it
                  has to be done online, but would be suitable for votes
                  in online communities etc. It's approval voting but
                  where the current scores are visible and where a voter
                  can change their vote as much as they like until the
                  deadline. But because co-ordinated factions might try
                  to mess with this by voting or changing their vote at
                  the last minute, I think a non-deterministic end point
                  might be necessary. So you might have a week of voting
                  plus and end section that has a half life of one hour.
                  There may be a sense in which this fails participation
                  though because with perfect game theoretical voting it
                  should become Condorcet. However, this is likely to be
                  from other voters' reactions to your presence in the
                  voting procedure rather than you voting against your
                  own interests. This is a fairly nebulous failure and
                  one I could easily tolerate for a clean results table
                  and a transparent and simple method. Also it might in
                  practice not always elect the Condorcet winner if the
                  Condorcet winner isn't particularly liked. E.g.</div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false"><br>
                </div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false">49 voters:
                  A>>C>B</div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false">49 voters:
                  B>>C>A</div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false">2 voters: C>A>B</div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false"><br>
                </div>
                <div
                  class="ydpe5cc3871yiv4812049359moz-forward-container"
                  dir="ltr" data-setdir="false">C (the Condorcet winner)
                  is unlikely to get off the ground in the first place
                  in this case.<br clear="none">
                  <br clear="none">
                  <blockquote type="cite">
                    <div dir="ltr"><br clear="none">
                    </div>
                    >>Two of the main ones I tend to look out for
                    are monotonicity and independence of clones because
                    they are obviously things we would want and they
                    don't >>seem to be too restrictive in terms of
                    methods they allow</blockquote>
                  <br clear="none">
                  >STAR fails both of them, "badly".<br clear="none">
                  <p dir="ltr" data-setdir="false">>STAR is obviously
                    garbage and a strategy farce,  as I'll explain later
                    on EM.<br clear="none">
                    <br clear="none">
                  </p>
                  <div dir="ltr" data-setdir="false">I look forward to
                    seeing it. As I say, I'm not a fan of STAR, but I am
                    still interested in seeing how it stands up to
                    scrutiny given that it has a following. (Actually
                    I'm not aware of how STAR fails monotonicity. I was
                    under the impression that it passed.)</div>
                  <div dir="ltr" data-setdir="false"><br>
                  </div>
                  <div dir="ltr" data-setdir="false">Toby</div>
                </div>
              </div>
            </div>
            <div class="ydpe5cc3871yqt5236108305"
              id="ydpe5cc3871yqtfd07461"><br clear="none">
            </div>
          </div>
        </div>
      </div>
    </blockquote>
  </body>
</html>