<html>
  <head>
    <meta http-equiv="content-type" content="text/html; charset=UTF-8">
  </head>
  <body dir="auto">
    <div><span></span></div>
    <div><br>
      <span></span></div>
    <div><br>
    </div>
    <div>Looking at election method as a purely mathematical problem,
      the objection to existing voting method is that it lacks a
      complete scale of measurement of candidate support, positive and
      negative. This is achievd by making an exclusion count the polar
      opposite of an election count, on the same continuum. The zero
      point in the middle is the zero surplus votes of just elected
      candidates. Or alternatively the zero deficit votes of just not
      unelected candidates.</div>
    <div>Once youve got this bipolar (or indeed binomial) count youve
      got one complete dimension, a basic standard of scientific
      measurement.</div>
    <div>(It's possible to go onto more than one dimension, as used in
      natural science.)</div>
    <div><br>
    </div>
    <div>Richard Lung.<br>
    </div>
    <div><br>
    </div>
    <div><br>
    </div>
    <div><br>
      <span></span><br>
      <span>Dear Richard Fobes, the VoteFair guy, and all at election
        methods,</span><br>
      <span></span><br>
      <span>Personal family misfortunes prevent me from giving a proper
        reply.</span><br>
      <span>I believe I say something about Kemeny in an appendix to my
        Smashwords free ebook, </span><br>
      <span>FAB STV: Four Averages Binomial Single Transferable Vote.</span></div>
    <div><font style="caret-color: rgb(0, 0, 0); background-color:
        rgba(255, 255, 255, 0);" color="#000000"><a
          href="https://www.smashwords.com/books/view/806030"
          style="caret-color: rgb(0, 0, 0); background-color: rgba(255,
          255, 255, 0);" moz-do-not-send="true">https://www.smashwords.com/books/view/806030</a></font></div>
    <div><br>
    </div>
    <div>The system is fully described in the second part, but there are
      plenty of summaries, from simple list of attributes, to summary
      convenient for those familiar with Meek method.</div>
    <div><br>
    </div>
    <div>I may mention what the system looks like from the voters point
      of view. It could be any preference voting ballot. But it counts
      differently. Last preferences help as much to exclude candidates,
      as first preferences help to elect candidates. That is to say it
      is a binomial count. </div>
    <div>It is not necessary to fill in all the preferences. Vacant
      preferences count towards a NOTA quota, leaving a seat unfilled.</div>
    <div><br>
    </div>
    <div>Circumstances permitting, I hope to say more about your post
      and others.</div>
    <div><br>
    </div>
    <div>Regards</div>
    <div>Richard Lung.</div>
    <div><br>
    </div>
    <div><br>
      <span></span><br>
      <span>On 28 Jul 2021, at 4:43 am, Richard, the VoteFair guy <<a
          href="mailto:electionmethods@votefair.org"
          moz-do-not-send="true">electionmethods@votefair.org</a>>
        wrote:</span><br>
      <span></span><br>
      <span>On 7/25/2021 2:00 PM, Richard Lung wrote:</span><br>
      <blockquote type="cite"><span>... eliminating candidates, during
          the count, loses voting information, before the count is over.
          ...</span><br>
      </blockquote>
      <span></span><br>
      <span>My favorite way to count ranked-choice ballots for a
        single-winner election is the Condorcet-Kemeny method.</span><br>
      <span></span><br>
      <span>Not only does it not eliminate candidates one at a time, it
        also does not identify the first-place winner as a first step,
        and the second-place winner as a second step, etc.  Instead it
        isn't finished until the entire sequence from most popular,
        second-most popular, and so on down to least popular has been
        determined.</span><br>
      <span></span><br>
      <blockquote type="cite"><span>I repeat, in case you missed it,
          science demands one truth (to aspire to) not two. Therefore an
          election count and an exclusion/elimination count must be
          symmetrical. Call it symmetrical count requirement. But that
          is a binomial count.</span><br>
      </blockquote>
      <span></span><br>
      <span>The Condorcet-Kemeny method is symmetrical.</span><br>
      <span></span><br>
      <blockquote type="cite"><span>Whereas, FAB STV is the whole
          caboodle perhaps relevant to data mining. This system is
          monotonic, not vulnerable to strategic shuffling the
          preference orders. It avoids premature exclusion, and indeed
          premature election! It entirely avoids later harm, not just
          for transfer of surplus preferences.</span><br>
      </blockquote>
      <blockquote type="cite"><span>It meets the Laplace condition of
          weighting a whole range of preferences in order of importance,
          unlike Condorcet pairing, whether or not the pairs are
          weighted in relative importance.</span><br>
      </blockquote>
      <blockquote type="cite"><span>FAB STV recognises elections as
          statistical estimates of representation, and employs up to
          four averages to maximise accuracy.</span><br>
      </blockquote>
      <span></span><br>
      <span>I have read your earlier posts but I don't recall seeing a
        description of your FAB STV method.</span><br>
      <span></span><br>
      <span>I looked at Electowiki but it's not there (under that name).</span><br>
      <span></span><br>
      <span>I'd be happy to look at FAB STV if you can point me to a
        definition of the method.</span><br>
      <span></span><br>
      <blockquote type="cite"><span>It accepts the "Impossibility" of a
          deterministic election result, and moves on. I beseech you all
          to do the same!</span><br>
      </blockquote>
      <span></span><br>
      <span>Remember that the RCIPE method (version 1 or 2) is a
        stepping stone that allows reaching better methods such as the
        Condorcet-Kemeny method, and beyond to PR methods.</span><br>
      <span></span><br>
      <span>Regarding the beyond part, my VoteFair Ranking system
        includes a two-seat kind of "STV," and that can be extended to a
        higher number of seats. That's why I'm curious to read a
        definition of your FAB STV counting method.</span><br>
      <span></span><br>
      <span>To put things into perspective, consider a metaphor. IRV is
        like a tricycle for tots who aren't yet ready for a bicycle. The
        Condorcet methods are like a bicycle, very useful in many
        situations, but they can't do everything. RCIPE is like training
        wheels for the bicycle. When enough voters learn how to vote
        using ranked-choice ballots and RCIPE counting then they will be
        ready to move on to Condorcet methods and beyond.</span><br>
      <span></span><br>
      <span>Richard Lung, thank you for your thoughts.</span><br>
      <span></span><br>
      <span>Richard Fobes</span><br>
      <span>The VoteFair guy</span><br>
      <span></span><br>
      <blockquote type="cite"><span>On 7/25/2021 2:00 PM, Richard Lung
          wrote:</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>A few comments from Richard Lung
          (not the VoteFair guy, (who is not to be confused, if I
          remember rightly, with Santucci, the vote guy!).</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>As pointed out to Susan Simmons,
          which she acknowledged, eliminating candidates, during the
          count, loses voting information, before the count is over. It
          is not necessary with a binomial count, unlike all existing
          methods (uninomial counts) which employ elimination as an
          afterthought to an essentially uninomial election count.</span><br>
      </blockquote>
      <blockquote type="cite"><span>I repeat, in case you missed it,
          science demands one truth (to aspire to) not two. Therefore an
          election count and an exclusion/elimination count must be
          symmetrical. Call it symmetrical count requirement. But that
          is a binomial count.</span><br>
      </blockquote>
      <blockquote type="cite"><span>And a binomial count indeed does
          imply higher order counts, governed by the binomial theorem.
          But a simple coherent first order binomial count should be
          sufficient for democratic representation.</span><br>
      </blockquote>
      <blockquote type="cite"><span>Whereas, FAB STV is the whole
          caboodle perhaps relevant to data mining. This system is
          monotonic, not vulnerable to strategic shuffling the
          preference orders. It avoids premature exclusion, and indeed
          premature election! It entirely avoids later harm, not just
          for transfer of surplus preferences.</span><br>
      </blockquote>
      <blockquote type="cite"><span>It meets the Laplace condition of
          weighting a whole range of preferences in order of importance,
          unlike Condorcet pairing, whether or not the pairs are
          weighted in relative importance.</span><br>
      </blockquote>
      <blockquote type="cite"><span>FAB STV recognises elections as
          statistical estimates of representation, and employs up to
          four averages to maximise accuracy. It accepts the
          "Impossibility" of a deterministic election result, and moves
          on. I beseech you all to do the same!</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>Yours sincerely,</span><br>
      </blockquote>
      <blockquote type="cite"><span>Richard Lung.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>On 25 Jul 2021, at 4:30 pm, Richard,
          the VoteFair guy <<a
            href="mailto:electionmethods@votefair.org"
            moz-do-not-send="true">electionmethods@votefair.org</a>>
          wrote:</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>On 7/24/2021 2:19 PM, Kristofer
          Munsterhjelm wrote:</span><br>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>On 7/22/21 5:31 PM, VoteFair
              wrote:</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>On 7/22/2021 6:04 AM, Kristofer
              Munsterhjelm wrote:</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>How about this?</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span></span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>- Eliminate the candidate with
                the least number of winning subgroups.</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>- If there is a tie, break
                that tie by IRV.</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>...</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span></span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>Isn't the first step basically
              Copeland's method?</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>No, because there's no elimination
            in Copeland (and it doesn't pass</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>LIIA). It would just elect the
            candidate/s with the most</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>winning subgroups.</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>I see you're right, of course.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>I admit your suggestion is clever
          because it includes Condorcet loser elimination.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>Yet I'm sure lots of non-math-savvy
          voters will not trust that the candidate with the least number
          of wins is not always the least popular.  I too share that
          lack of trust.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>Keep in mind that lots of voter
          don't trust the idea that the winner of all the pairwise
          contests is always the most popular.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>But then clone independence is not
            important after all because the</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>methods are ugly. I can't quite
            determine whether clone independence is</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>important or not.</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>It's important that the failure rate
          is small. But it doesn't need to be zero.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>That's true. You implicitly need
            some kind of valuation of the different</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>failure rates. For instance, if
            you want LNHarm and LNHelp, you have to</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>give up either monotonicity or
            mutual majority. Which it's going to be</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>depends on what values you place
            on the different criteria.</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>I'm not concerned about
          monotonicity, LNHarm, LNHelp or any other On 7/24/2021 2:19
          PM, Kristofer Munsterhjelm wrote:</span><br>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>On 7/22/21 5:31 PM, VoteFair
              wrote:</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>On 7/22/2021 6:04 AM, Kristofer
              Munsterhjelm wrote:</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>How about this?</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span></span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>- Eliminate the candidate with
                the least number of winning subgroups.</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>- If there is a tie, break
                that tie by IRV.</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>...</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span></span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>Isn't the first step basically
              Copeland's method?</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>No, because there's no elimination
            in Copeland (and it doesn't pass</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>LIIA). It would just elect the
            candidate/s with the most</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>winning subgroups.</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>I see you're right, of course.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>I admit your suggestion is clever
          because it includes Condorcet loser elimination.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>Yet I'm sure lots of non-math-savvy
          voters will not trust that the candidate with the least number
          of wins is not always the least popular.  I too share that
          lack of trust.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>Keep in mind that lots of voter
          don't trust the idea that the winner of all the pairwise
          contests is always the most popular.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>But then clone independence is not
            important after all because the</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>methods are ugly. I can't quite
            determine whether clone independence is</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>important or not.</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>It's important that the failure rate
          is small. But it doesn't need to be zero.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>That's true. You implicitly need
            some kind of valuation of the different</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>failure rates. For instance, if
            you want LNHarm and LNHelp, you have to</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>give up either monotonicity or
            mutual majority. Which it's going to be</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>depends on what values you place
            on the different criteria.</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>I'm not concerned about
          monotonicity, LNHarm, LNHelp or any other failures that are
          difficult to exploit. I'm much more concerned about
          exploitable failures.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>Admittedly, as a fan of
          Condorcet-Kemeny, I favor looking deep into the ballots, and I
          favor ways of "sorting" that basically move the biggest
          pairwise counts into one half of the usual matrix while moving
          the smallest pairwise counts into the other half, where the
          dividing line is the diagonal where candidates are paired with
          themselves.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>Or to put it differently: if the
            method insists on a zero failure rate</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>for Condorcet loser, why shouldn't
            it insist on a zero failure rate for</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>Condorcet winner, say? And,
            equivalently, if "merely a low rate of</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>failure" is good enough for the
            Condorcet criterion (or say, clone</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>independence), why is it not good
            enough for Condorcet loser?</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>I admit I'm intentionally avoiding a
          zero failure rate for Condorcet winner because that makes the
          method into a Condorcet method, and those have been vilified
          (portrayed as evil) by the FairVote organization, and to some
          extent by STAR fans.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>Plus, just as a voter is not likely
          to trust that the candidate with the fewest wins is least
          popular, they aren't likely to trust that the candidate who
          wins all the pairwise matches is most popular.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>So at this point I'm still happy
          with eliminating the Condorcet loser as the top priority and
          otherwise eliminating the candidate who has the smallest
          pairwise support count (which basically counts how many
          remaining candidates are ranked below the candidate being
          scored).</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>At this point I continue to be open
          to suggestions for something better, but that window of time
          is closing very soon.</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>Again, thank you Kristofer for your
          wise feedback!</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span>Richard Fobes</span><br>
      </blockquote>
      <blockquote type="cite"><span>The VoteFair guy</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>On 7/24/2021 2:19 PM, Kristofer
              Munsterhjelm wrote:</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>On 7/22/21 5:31 PM, VoteFair
              wrote:</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>On 7/22/2021 6:04 AM,
                Kristofer Munsterhjelm wrote:</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>How about this?</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span></span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>- Eliminate the candidate with
                the least number of winning subgroups.</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>- If there is a tie, break
                that tie by IRV.</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>...</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span></span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>Isn't the first step basically
              Copeland's method?</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>No, because there's no elimination
            in Copeland (and it doesn't pass</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>LIIA). It would just elect the
            candidate/s with the most winning subgroups.</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>That's an ugly "method" that
              fails to look beneath the surface.</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span></span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>IRV also fails to look beneath
              the surface, which is why it too is an</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>"ugly" method.</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>That leads me to wonder which is
            the case.</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>You said you couldn't replace the
            IRV tiebreaker with minmax elimination</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>because IRV is cloneproof and
            minmax is not -- that clone independence</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>was important because it "protects
            against money-based vote splitting</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>tactics". So I found something
            that invokes IRV's clone independence</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>more often.</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>But then clone independence is not
            important after all because the</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>methods are ugly. I can't quite
            determine whether clone independence is</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>important or not.</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite"><span>But again, the ungrouped
                mechanic is not cloneproof.</span><br>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span></span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>Being cloneproof is not a goal.
              The goal is to have a very small</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>failure rate for clone
              independence.</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>Then you could check the
            alternatives by that metric. A method seeming</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>ugly may not necessarily have any
            bearing on the rates of failure.</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>Also, electing the Condorcet
              winner is not a goal. The goal is to have</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>a very small Condorcet criteria
              failure rate.</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span></span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>To repeat my concern, attempting
              to get a zero failure rate will cause</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>other kinds of failure rates to
              increase.</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>That's true. You implicitly need
            some kind of valuation of the different</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>failure rates. For instance, if
            you want LNHarm and LNHelp, you have to</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>give up either monotonicity or
            mutual majority. Which it's going to be</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>depends on what values you place
            on the different criteria.</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>The same would hold for rates. Say
            you want to find the method that</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>minimizes w * x, where x is the
            rates of each failure type</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>(monotonicity, vote splitting,
            teaming, crowding, favorite betrayal...).</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>Then the weights of the w vector
            provide a measure of indifference: how</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>much of failure type 1 is an
            acceptable trade for one unit of failure</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>type 2?</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>Or to put it differently: if the
            method insists on a zero failure rate</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>for Condorcet loser, why shouldn't
            it insist on a zero failure rate for</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>Condorcet winner, say? And,
            equivalently, if "merely a low rate of</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>failure" is good enough for the
            Condorcet criterion (or say, clone</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>independence), why is it not good
            enough for Condorcet loser?</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>I'm still willing to consider
              improvements, but it needs to find a</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>balance between what voters can
              understand -- both through an animated</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>video and through words -- and
              what yields low failure rates.</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span></span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>Again, thank you Kristofer for
              applying your clear understanding to</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite"><span>this revision from RCIPE 1 to
              RCIPE 2.</span><br>
          </blockquote>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>You're welcome :-)</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span></span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite">
        <blockquote type="cite"><span>-km</span><br>
        </blockquote>
      </blockquote>
      <blockquote type="cite"><span>----</span><br>
      </blockquote>
      <blockquote type="cite"><span>Election-Methods mailing list - see
          <a href="https://electorama.com/em" moz-do-not-send="true">https://electorama.com/em</a>
          for list info</span><br>
      </blockquote>
      <blockquote type="cite"><span></span><br>
      </blockquote>
    </div>
  </body>
</html>