Loading ...
Sorry, an error occurred while loading the content.

purpose of group and how to use it

Expand Messages
  • wds@math.temple.edu
    THE PURPOSE of the RangeVoting@yahoo.com group is to serve as a communication and archival medium (bulletin board with archival storage and search capability)
    Message 1 of 9 , Aug 4, 2005
    • 0 Attachment
      THE PURPOSE of the RangeVoting@... group is to serve as
      a communication and archival medium (bulletin board with archival storage and
      search capability) for people interested in range voting and related subjects.


      OUR TWO MAIN GOALS are
      (1) to create and disseminate knowledge about range voting.

      (2) to actually lobby (and organize, investigate, and explain the lobbying and activism)
      to cause various governments and organizations to adopt range voting as
      their method of performing multi-candidate single-winner elections.


      HOW TO MAKE RANGE VOTING HAPPEN
      As of August 2005, I believe that the three most important and productive paths to take are:

      (1) Push to make either the Democratic or Republican (USA) party (or both) adopt
      range voting in their Iowa 2008 "caucuses". Whichever party does this will be likely to
      benefit greatly by selecting a better presidential candidate, and also by
      getting a lot of free publicity about the fact that they are reforming the voting system
      in order to benefit us all. If the opposing party fails to do the same then your
      party will have a large moral edge, a publicity advantage, as well as a likely
      quality-of-candidate advantage that will help in the 2008 presidential contest.
      This is a move that will cost you essentially nothing and will lead to advantage.

      In contrast, plurality elections among 10 candidates (which is what happened to
      the Democrats in Iowa 2004) tend to lead to poor choices dominated by short-term
      strategic considerations instead of long-term quality considerations.
      (The Democrats consequently ended up nominating John Kerry, who seems to have
      been a poor choice for them since he last the election to Bush.) Don't be that
      stupid again!

      (2) Push to get range voting implemented in nascent democracies such as (?) Iraq
      or China. Countries that institute range voting for their single winner elections
      and votes will enjoy improved government compared to countries emplying
      single-winner plurality voting.

      (3) Adopt range voting for multi-alternative 1-winner votes inside organizations such as
      corporations (corporations, while not usually considered to be a democracy, often use such
      votes). The result will be better decisions resulting in increased profits, at essentially
      no cost to you. It is absolutely irresponsible for corporations to employ,
      say, Borda voting or plurality voting to make, e.g, hiring decisions.


      WHAT RangeVoting@... ENABLES YOU TO DO, and HOW TO DO IT:
      (1) Start by pointing your browser to http://groups.yahoo.com/group/RangeVoting
      Now by following various hyperlinks and instructions, you can:
      (2) become a MEMBER. (Click "sign up." You may also want to check the
      "membership wizard" at http://groups.yahoo.com/memwiz.)
      (3) READ and text-SEARCH archives of all previously posted messages.
      This is highly recommended since that way you can avoid asking questions which have
      already been answered and you can rapidly bring yourself up to the leading edge.
      (Click "messages" and enter text in "search" box.
      Both subjects, authors, and message texts are searched.)
      (4) examine the HYPERLINKS list. (Click "links.")
      (5) examine the CALENDAR. (Click "calendar.")
      (6) POST your own message. (Send email to RangeVoting@yahoogroups.com.)
      Once you are a MEMBER, you can:
      (7) Get all posts to RangeVoting delivered to you automatically either immediately,
      as a "daily digest," or put on hold until you get back from vacation.
      (8) examine and download FILES and PHOTOS. (Click "files.")
      You presently cannot post files and photos except with the aid of a moderator.
      (9) participate in POLLS either by voting in them or creating them (click "polls.")
      (10) view profiles of other members. (click "members.")
      (11) promote the group by downloading nifty html widgets to put on your web page that
      allow anybody viewing your web page to easily sign up for range voting.
      (click "promote.")
      (12) invite other potential members. (click "invite.")
      (13) unsubscribe. (RangeVoting-unsubscribe@yahoogroups.com, and see
      http://help.yahoo.com/help/us/groups/groups-32.html)
      Extra powers come if you are a MODERATOR. Initially the only moderator
      is the creator of this group, Warren D. Smith. However, the moderators may
      decide at any point to create other moderators, sort of like the spawning of alien
      monsters from eggs.

      -wds
    • abdlomax
      ... Extra powers come if you are a MODERATOR. Initially the only moderator is the creator of this group, Warren D. Smith. However, the moderators may decide
      Message 2 of 9 , Aug 4, 2005
      • 0 Attachment
        --- In RangeVoting@yahoogroups.com, wds@m... wrote:

        "Extra powers come if you are a MODERATOR. Initially the only
        moderator is the creator of this group, Warren D. Smith. However, the
        moderators may decide at any point to create other moderators, sort of
        like the spawning of alien monsters from eggs."

        As Mr. Smith knows, my concern is group process. That concern got me
        booted from the ApprovalVoting list, a sad and silly outcome, I think.
        The AV list is a good example of how having just a single moderator,
        beyond the initial formative stages of a list, is a bad idea. In this
        case, the moderator allowed his personal reaction to my posts (both on
        and off the list) to carry him away, so that he put me on moderation
        and started deleting my posts without notice, then, when he realized
        that this was going to look really bad, he started rejecting posts
        instead. He was about to take me *off* moderation, but when I refused
        to agree to follow vague rules created just for me, he, again without
        warning, set my privileges to Not Allowed to Post, which automatically
        rejects them. (There were three pending posts at that time, which were
        thus rejected. They were all relevant to Approval Voting and one of
        them was actually a direct response to the appeal of the President of
        CAV for help with setting up a way for a larger group to be able to
        create content for the CAV web site, where I offered that help, and
        also pointed out that a usable site already existed: the wiki
        av.beyondpolitics.org.

        In a list with more than one moderator, it really takes a consensus
        among the moderators to more than temporarily prevent a member from
        posting, for any other moderator can reverse the action.

        Likewise, any unmoderated member of a list can bypass moderation, if
        the member considers a post submitted to him or her relevant and
        appropriate. It is simply passed on as "mail I received."

        Yahoogroups lists have one single *owner*, who typically has the power
        to create or remove moderator privileges, and usually would be the
        only one who can trash the list. Moderators have powers as set by the
        list owner.

        Free Association / Delegable Proxy (FA/DP) is an set of organizational
        concepts that apply in general to voluntary peer associations. Such as
        an informal list like this. An FA/DP mailing list would start as a
        simple unmoderated list. However, at some point a proxy list would be
        made, which is simply a list of showing an ID for each member,
        together with a single ID of another member who has been assigned
        proxy by the first. If any. One of the things that such a list does is
        to measure a level of trust. And it makes things like selecting
        moderators who are broadly trusted easy. So new moderators might be
        chosen from among those who were broadly trusted. If the list owner
        wants to give up the position, I'd consider that an elective position,
        and Approval might be the exact best method....

        When the list gets really large, having it unmoderated by default can
        become a burden. Again, in an FA/DP organization, a list would
        presumably start to define categories of members: default membership
        would now not include the right to post without moderator approval.
        And members with a certain level of trust (not large, necessarily, but
        such members should represent at least a few general members) would be
        allowed to post directly without moderation.

        All this is looking way ahead: while we might hope for a level of
        success that would require examining these issues soon, I rather doubt
        it. The AV list has about 86 members and it is only now seeing some
        feeling that posting should be restricted in some way, at least to
        prevent the cranks (like me) from filling up the bandwidth.

        Still, having more than one moderator, initially by choice of the
        existing moderators, is really a good idea, for moderators can be
        absent for periods of time. The moderator of the AV list informed me
        that I was banned and he was going away, so if I reconsidered my
        intransigence, he might change his mind when he comes back. Of course,
        I wasn't being intransigent, I was merely not readily agreeing to
        what he had offered me (promise to obey his ad hoc rules), as
        supposedly something better than what I was proposing. Which was that
        if posts were rejected, and the writer did consider it relevant to AV,
        a very brief precis of the post with a link to the metalist
        AVFA@yahoogroups.com would be allowed, at least occasionally (it might
        cover several messages). So readers could decide for themselves; this
        proposal took the wind out of the sails of his objections to my
        postings, I'd have thought. As it would with his objections to Mr.
        Smith's postings. Those summary posts would have been at least as
        relevant as the average AV post, and would not have taken up undue
        bandwidth....

        And by the way, AVFA@yahoogroups.com does exist for discussing
        Approval Voting with a much broader definition of what is relevant. As
        the list for the Approval Voting Free Association, it is technically
        open to people who absolutely oppose Approval Voting; however, the
        list is not a free-for-all, and moderation would be available as
        needed. The point, though, is that Range Voting would be especially
        welcome there, given the close relationship with Approval, but also
        discussions of AV vs. IRV, etc. Hopefully in a collegial atmosphere,
        but, of course, that cannot be guaranteed, merely encouraged.

        Mr. Smith, you should announce this Range Voting list on the AV list,
        if you are permitted, and, if not, you should announce it on the AVFA
        list.
      • WarrenS
        Laraki claimed he had to fly back to France and hence virtually no time to talk with me. Only was able to talk about 5 mintes and nothing substantive said.
        Message 3 of 9 , Jul 12 12:29 PM
        • 0 Attachment
          Laraki claimed he had to fly back to France and hence
          virtually no time to talk with me. Only was able to talk about 5 mintes
          and nothing substantive said. Laraki claimed he'd not been replying to my emails because he "liked to think, not answer."
        • WarrenS
          Dan S Felsenthal: Review of paradoxes afflicting various voting procedures where one out of m candidates (m ≥ 2) must be elected,
          Message 4 of 9 , Jul 12 1:40 PM
          • 0 Attachment
            Dan S Felsenthal:
            Review of paradoxes afflicting various
            voting procedures where one out of m
            candidates (m ≥ 2) must be elected,
            http://eprints.lse.ac.uk/27685/1/Review_of_Paradoxes_Afflicting_Various_Voting_Procedures_(LSERO).pdf

            and a later and different version of this was
            published as a book chapter in
            DS Felsenthal & M Machover (editors): Electoral systems, Springer.
            I am mostly going by the book's version.

            Anyhow. Felsenthal defines 15 voting "paradoxes" and also 17 voting systems. He then constructs a chart saying which system is afflicted by which paradox. (None of this is new. All chart entries were previously known.) For most but not all of the entries in the table, he gives election examples to prove it.

            Some kinds of paradoxes, Felsenthal considers severe; other kinds bother him less.

            So anyhow, here is an abbreviated version of Felsenthal's chart:

            xxxxxxxxxxxxxxxxx...Range...MJ...Borda...IRV...Coombs...Bucklin...Cond..Plur
            #severe violations:...3......3.....1.......1.....1.........1........0....2
            #non-severe viltns:...4......7.....5.......9.....9.........9........7....4

            where
            MJ="Majority Judgment" median-score based procedure by Balinski+Laraki,
            Borda=candidate ranked M spots above bottom gets score M, greatest total
            score wins,
            Plur=simple "name one candidate" voting,
            IRV=instant runoff voting (repeatedly eliminate candidate with
            fewest top-rankings until only 1 survives),
            Coombs is similar to IRV except we repeatedly eliminate the
            candidate with the fewest bottom-rankings,
            and
            "Cond" is any of 3 different Condorcet procedures, namely Black,
            Copeland, and Kemeny. The Black variant is the simplest: it is to
            elect the Condorcet winner if one exists, otherwise elect Borda winner.
            (Felsenthal also discusses other Condorcet procedures all of which he considers
            worse than these three.)

            So, e.g, according to his chart, Borda and the three mentioned Condorcet
            methods are the best, and both are superior to range voting. Indeed,
            according to his chart even plain plurality voting is superior to range voting.

            Sigh.

            So now, let us examine the details.
            The three "severe" violations range voting suffers are
            1.Absolute majority
            2.Condorcet loser
            3.Absolute loser
            all of which are simultaneously
            demonstrated by the following example:

            51 voters: Hitler=5, Truman=4.
            49 voters: Truman=9, Hitler=0.

            Range voting elects Truman, which according to Felsenthal
            is a "severe paradox" because it evidently is his view that
            Hitler should have been elected.

            Of course, a different reading of the situation would be
            that all three of those logical criteria actually should
            not be demanded! Felsenthal was simply wrong to label
            the violation of those criteria a "paradox" much less
            brand it "severe." But of course Felsenthal does not
            even mention that. It does not even occur to him.
            Another thing to consider is the fact that all three
            of these violations come with asterisks:
            1.If an absolute majority wants X, they can force X's election with range voting.
            2.In the real world where many voters are strategic, there is good reason to suspect, backed up by computer simulations, range will elect condorcet winners in practice more often than condorcet methods do(!), see
            http://rangevoting.org/AppCW.html
            3.If an absolute majority does not want X, they can force X to lose
            with range voting.

            In view of these 3 asterisks, calling these "severe" was ludicrous.
            But Felsenthal does not even mention the asterisks.

            The 4 non-severe paradoxes Range Voting suffers from (says Felsenthal) are
            1.Condorcet "intransitive" cycle can exist.
            2.Strategic voting: there can be situations where voter is
            better off lying in her vote than honest voting.
            (Felsenthal claims every method he charts suffers both problems 1 & 2.)
            3.Condorcet winner can exist but not be elected by range voting.
            4.Truncation -- it can be strategically superior for some voter
            to leave some candidate unscored.

            OK, let us reply to these.
            Re 1, the condorcet cycle problem is a matter of self-inconsistency.
            That is why it is called a "paradox."
            Range voting is not inconsistent, i.e. it can NEVER say that A>B>C>A
            where ">" means "has greater mean score." This differs
            from (say) Black's method, where Black can easily say that A
            would beat B if the other rankings (for C) were erased
            from all ballots. (And etc for B vs C, and C vs A.)

            Therefore, range voting does not suffer a "paradox" of this kind,
            contrary to Felsenthal. Of course, by redefining the
            word "paradox" to mean "I, Dan Felksenthal do not like it"
            that difficulty is avoided. However, using
            the dictionary definition, this is no paradox, and indeed t
            his is an advantage of range voting versus every other method
            on his chart except MJ. Felsenthal fails to comprehend that.
            Also, although a consideration of voting systems using
            the "Felsenthal does not like it" subjective metric
            perhaps is also of some interest, the fact that Felsenthal
            prefers that Hitler win in above scenario, means his judgment is poor.

            Re 2, Felsenthal LSE version QUOTE
            Ceteris paribus, a voter may obtain a preferred outcome if he votes strategically, i.e, not according to his true
            preferences. All known voting procedures suffer from this paradox.
            END QUOTE.

            Felsenthal is actually correct that under that definition
            range voting does suffer this problem,
            but gives no proof and I suspect based on his general level of
            ignorance and incompetence that he is incapable of producing a proof.

            Re 3, see above "severe" issue #2, same response applies.

            Re 4, Felsenthal gives this example to "prove" his contention:

            CANDIDATE....V1....V2....V3....V4...V5...V6...V7...mean score
            X.........1.....1....1.....10....5....4....7.....4.143
            Y.........2.....2....2......3....8....5....8.....4.286

            In this situation (Felsenthal says) Y wins with range voting.
            Voter 4 (denoted V4) is unhappy about that. Voter 4 is better off
            NOT SCORING Y, because then Y would have won. Q.E.D.
            Unfortunately for Felsenthal, he's wrong. The rules
            of range voting he cites [namely rangevoting.org web page,
            and also the rules he himself describes in his section 3.2.2.5]
            say highest average wins. If V4 leaves Y unscored, then X still wins,
            indeed by more margin. Felsenthal invents on the spot the new rule,
            which he had not stated when describing the rules of range voting, that
            "no score" counts the same as "0 score." Under that new rule,
            Y would win.

            Similarly, if I invented new rules on the spot when inventing paradoxical situations, I could make any voting method fall to any paradox I wanted.

            OK, so let's tote up the score. Felsenthal makes 2 mistakes
            about range voting chart entries. He also
            misunderstands/misuses/redefines the
            meaning of "paradox" which in an article about paradoxes,
            is rather bad. He also leaves many other paradoxes out of his
            considerations entirely, whicjh if they were included would change
            the picture.

            He states the following in his (LSE version) paper:
            QUOTE
            Among the 17 (deterministic) voting procedures analysed in this paper, the Condorcet-consistent procedures proposed by Copeland (1951) and by Kemeny (1969) seem to me to be the most
            desirable from a social-choice perspective for electing one out of several candidates.
            END QUOTE

            OK, that sentence reveals that Felsenthal is an idiot.

            First of all, with Kemeny it is NP-hard to determine the winner, meaning
            in a large election we could not guarantee we would be able to determine the winner even with all computers in world running for years. That means Kemeny is
            not "the most desirable" it is "unacceptable for large elections."

            Second, with Copeland it is common for elections to be exact ties
            even with millions of voters. You can confirm this with
            a simple computer simulation, which obviously Felsenthal has never done.

            The job of a voting method, is to produce one winner. Kemeny and Copeland
            fail to perform that job. The question of what "paradoxes" that winner
            may suffer from, is secondary. First you have to DO your job, THEN we reckon
            whether it was a good job. If you never do your job, then you do not
            even get to enter the contest.

            I have told Felsenthal and Machover all this well before Felsenthal
            wrote this paper.
            He ignored me and wrote the paper anyway, never even mentioning these
            disqualifying problems with Kemeny and Copeland.
            So in addition to writing a paper containing nothing new, he also
            published mistakes.
          • WarrenS
            So for example, suppose Felsenthal had added to his list of paradoxes these additional ones his paper failed to consider: 1. partitioning into districts
            Message 5 of 9 , Jul 12 3:08 PM
            • 0 Attachment
              So for example, suppose Felsenthal had added to his list of "paradoxes" these additional
              ones his paper failed to consider:

              1. partitioning into districts paradox I:
              Suppose a country is divided into districts and candidate A wins in every district. If A does not win in the whole country, that is a paradox.

              2.partitioning into districts paradox II:
              Suppose a country is divided into districts and candidate A places last in every district. If A does not place last in the whole country, that is a paradox.

              3. DH3 pathology. See http://www.rangevoting.org/DH3.html .
              (I would call this "severe", incidentally; reasons why explained there.)

              4. Reversal paradox: A is the winner. After all voters reverse their ballots (trying to determine the worst rather than best candidate using voting system) A is still the "winner."

              5. Generic untied: Consider every possible V-voter C-candidate election for every possible integer V=2,3,4,5,...,U. Assume the election is randomly selected from these possibilities and take the limit U-->infinity. If the probability this election is a perfect tie, is nonzero, that is a pathology.

              6. Feasible to compute winner: If Dan Felsenthal, given 3 years to do it, millions of
              dollars in funding, and all the assistants he can hire, can write a computer program
              for election system E which
              A. Given any number of votes up to 1 billion and candidates up to 200, as input
              B. will always compute the election-system-E winner
              C. will do it within 1 week of compute time (or less)
              then E is "computationally feasible." If not, then E is "unacceptable."

              7. Favorite-safe: if it can be strategically forced for a voter to lie about the identity of her favorite candidate, then that is a paradox.

              8. Clone-safe: if a "clone" of a candidate (rated almost identical to the original by every voter) leaves the race, that should not affect the winner (aside from possible replacement by a clone). If the winner changes, this is a paradox.

              9.Remove-loser safe: If some losing candidate X is found to be a criminal and ineligible to run, then the same ballots should still be useable to conduct an election with X removed, and should still elect the same winner. If the winner changes, that was a paradox.

              Further, suppose Felsenthal (whom I see is employed by London school of economics)
              considers it more desirable that 49 people get a million dollars worth of benefit each, then
              that 51 people get 1 dollar of benefit each. In that case, Felsenthal agrees that "majority winner" is not socially desirable. In that case, he agrees that demanding a voting method always obey "absolute majority", always obey "condorcet winner", always obey "condorcet loser" or always obey "absolute majority loser" are not demands he supports.

              In that case, he would agree to remove those so-called paradoxes from his paper, or to redefine them with improved wordings. I can help him with the latter if he needs it.

              OK, what would the new, improved-with-extra-criteria and with bad ones modified or removed, Felsenthal paper then conclude?

              Well golly. Gee whiz. Range voting obeys all 9 of my new
              criteria. Felsenthal's favored "Copeland Condorcet" and "Kemeny Condorcet"
              voting methods, each disobey 8 of the 9.
              That would change the totals to

              OLD TOTALS:
              RANGE VOTING: 3 severe plus 4 non-severe violations
              BLACK & COPELAND: each 0 severe and 7 non-severe

              AFTER CORRECT 1 or 2 ERRORS BY FELSENTHAL:
              RANGE VOTING: 3 severe plus 2 or 3 non-severe violations
              BLACK & COPELAND: each 0 severe and 7 non-severe

              AFTER ADD 9 NEW CRITERIA:
              RANGE VOTING: 3 severe plus 2 or 3 non-severe violations
              BLACK & COPELAND: each 0 severe and 7 non-severe and 8 new

              And the 4 removed critieria (or modified)? Range voting's failures of them no longer matter since removed, or it obeys the modified versions. All RV "severe" problems go away.

              And golly. Gee whiz. Range voting now is better, according to Felsenthal's own reasoning with extra juice added, than Kemeny and Copeland, his two self-described favorite methods.

              Indeed, Felsenthal's old chart had considered MJ the *worst* voting method, but the new as-revised chart would consider MJ "better than the methods Felsenthal called best."

              Amazing, isn't it. Complete reversal of Felsenthal conclusions by improving his paper.

              That means his paper was really bad.
              But yet, by the simple method of publishing it in a book edited by himself, it is now
              sitting out there in a recently published, allegedly state of the art, voting methods book.
            • WarrenS
              ... of range voting he cites [namely rangevoting.org web page, and also the rules he himself describes in his section 3.2.2.5] say highest average wins. If V4
              Message 6 of 9 , Jul 12 3:14 PM
              • 0 Attachment
                >Unfortunately for Felsenthal, he's wrong. The rules
                of range voting he cites [namely rangevoting.org web page,
                and also the rules he himself describes in his section 3.2.2.5]
                say highest average wins. If V4 leaves Y unscored, then X still wins,
                indeed by more margin. Felsenthal invents on the spot the new rule,
                which he had not stated when describing the rules of range voting, that
                "no score" counts the same as "0 score." Under that new rule,
                Y would win.

                --sorry, correction:
                swap "X" and "Y" in above paragraph. (I'm still correct, and Felsenthal still
                is wrong, but I typed the wrong letters.)
              • Markus Schulze
                Hallo, I have read the published version of this paper: Dan S. Felsenthal, Review of Paradoxes Afflicting Procedures for Electing a Single Candidate , eds.
                Message 7 of 9 , Jul 12 11:52 PM
                • 0 Attachment
                  Hallo,

                  I have read the published version of this paper:

                  Dan S. Felsenthal, "Review of Paradoxes Afflicting
                  Procedures for Electing a Single Candidate",
                  eds. Dan S. Felsenthal, Moshe Machover, "Electoral
                  Systems -- Paradoxes, Assumptions, and Procedures",
                  Springer-Verlag, pages 19-91, 2012,
                  DOI: 10.1007/978-3-642-20441-8_3

                  In this published version, he writes about the
                  Schulze method:

                  "I also do not list here two Condorcet-consistent
                  deterministic procedures proposed by Tideman (1987)
                  and by Schulze (2003) because I do not consider
                  satisfying (or violating) the independence-of-clones
                  property, which is the main reason why these two
                  procedures were proposed, to be associated with any
                  voting paradox. A phenomenon where candidate x is
                  more likely to be elected when two clone candidates,
                  y and y', exist, and where x is less likely to be
                  elected when, ceteris paribus, one of the clone
                  candidates withdraws, does not seem to me
                  surprising or counter-intuitive."

                  I guess that many readers (esp. many IRV supporters)
                  disagree with this.

                  Markus Schulze
                • WarrenS
                  ... --yes, the more I look at it, the more my mind is boggled by how bad the Felsenthal paper is. I unfortunately am going to have to prepare a web page
                  Message 8 of 9 , Jul 13 5:40 AM
                  • 0 Attachment
                    --- In RangeVoting@yahoogroups.com, Markus Schulze <markus.schulze@...> wrote:
                    >
                    > Hallo,
                    >
                    > I have read the published version of this paper:
                    >
                    > Dan S. Felsenthal, "Review of Paradoxes Afflicting
                    > Procedures for Electing a Single Candidate",
                    > eds. Dan S. Felsenthal, Moshe Machover, "Electoral
                    > Systems -- Paradoxes, Assumptions, and Procedures",
                    > Springer-Verlag, pages 19-91, 2012,
                    > DOI: 10.1007/978-3-642-20441-8_3
                    >
                    > In this published version, he writes about the
                    > Schulze method:
                    >
                    > "I also do not list here two Condorcet-consistent
                    > deterministic procedures proposed by Tideman (1987)
                    > and by Schulze (2003) because I do not consider
                    > satisfying (or violating) the independence-of-clones
                    > property, which is the main reason why these two
                    > procedures were proposed, to be associated with any
                    > voting paradox. A phenomenon where candidate x is
                    > more likely to be elected when two clone candidates,
                    > y and y', exist, and where x is less likely to be
                    > elected when, ceteris paribus, one of the clone
                    > candidates withdraws, does not seem to me
                    > surprising or counter-intuitive."
                    >
                    > I guess that many readers (esp. many IRV supporters)
                    > disagree with this.
                    >
                    > Markus Schulze

                    --yes, the more I look at it, the more my mind is boggled by how bad the Felsenthal paper is. I unfortunately am going to have to prepare a web page specifically to attack this paper.





                    >
                  • Abd ul-Rahman Lomax
                    This is long, but I think it gets into some fundamental issues, with a bit of chatter as well. ... Fascinating to see what shallow tripe is produced by
                    Message 9 of 9 , Jul 14 6:33 PM
                    • 0 Attachment
                      This is long, but I think it gets into some
                      fundamental issues, with a bit of chatter as well.

                      At 04:40 PM 7/12/2013, WarrenS wrote:
                      >Dan S Felsenthal:
                      >Review of paradoxes afflicting various
                      >voting procedures where one out of m
                      >candidates (m ≥ 2) must be elected,
                      >http://eprints.lse.ac.uk/27685/1/Review_of_Paradoxes_Afflicting_Various_Voting_Procedures_(LSERO).pdf

                      Fascinating to see what shallow tripe is produced
                      by academics who are actually paid.

                      The abstract:

                      >The paper surveys 17 deterministic electoral
                      >procedures for selecting one out of two or more
                      >candidates, as well as the susceptibility of
                      >each of these procedures to various paradoxes. A
                      >detailed appendix exemplifies the paradoxes to
                      >which each electoral procedure is susceptible.
                      >It is concluded that from the perspective of
                      >vulnerability to serious paradoxes, as well as in
                      >light of additional technical-administrative
                      >criteria, Copeland’s or Kemeny’s proposed
                      >procedures are the most desirable.

                      They do not seem to be aware that "election
                      paradoxes" are only paradoxes with respect to
                      naive intuition. They show no discrimination
                      between system failures that frustrate the
                      purpose of elections, and those which are only
                      formal. Thus, to them, the Majority Criterion
                      failure of Range Voting is *fatal*, while any
                      sane social choice process must ultimately accept
                      certain MC failures in an *initial poll*. They
                      have no concept of the importance of preference
                      *strength*. They have no concept of how voters
                      would *actually vote* in Range voting, where they
                      care about the election or non-election of a
                      candidate. Hence, with Range Voting, they show
                      the "majority" paradox, but the voting pattern
                      they use actually shows that the voters *don't
                      care* about the preference asserted.

                      I don't have the book, so I'm only commenting on
                      the paper. *There actually was a book published
                      by Springer,* the second-largest academic and
                      scientific publisher in the world, at least roughly.

                      From what motivated the writing of the paper:

                      >I therefore thought it would be well to
                      >supplement that report by reminding social choice
                      >theorists, political scientists, as well as
                      >commentators, policymakers and interested laymen –
                      >especially in the UK and the US – of the main
                      >social-choice properties by which voting
                      >procedures for the election of one out of two or
                      >more candidates ought to be assessed, and to
                      >list and exemplify the paradoxes afflicting these voting procedures.

                      Okay, great idea. By what "properties" ought
                      voting procedures be assessed? That is not the
                      title of the paper, rather, the papar focuses on
                      the paradoxes." I.e., who cares about the *actual
                      purpose* of voting procedures, let's look at
                      what's funny! Paradoxes are the foundations of
                      jokes, the unexpected. If we believe in rigid
                      standards, paradoxes expose what's "wrong."
                      Paradoxes abound where there are fixed and unexamined assumptions.

                      When we understand and recognize these, a
                      "paradox' becomes a truly funny joke, a
                      contradiction in *terms*, not in reality. I'll
                      simply assert, without proof, that there are no contradictions in reality.

                      >Thus this paper should be regarded as an updated
                      >review by which to assess from a socialchoice
                      >perspective the main properties of various known
                      >voting procedures for the election of
                      >a single candidate.
                      >Of the17 (deterministic) voting procedures
                      >analysed in this paper, the Condorcet-consistent
                      >procedures proposed by Copeland (1951) and by
                      >Kemeny (1969) seem to me to be the most
                      >desirable from a social-choice perspective for
                      >electing one out of several candidates.

                      Yet he never considers social choice theory. He's
                      using "social-choice perspective" to refer only
                      to behavior with regard to "voting system
                      criteria," a thoroughly bankrupt approach. No
                      social choice theorist, worthy of that epithet, would make this gross blunder.

                      What has happened is that political academics
                      have relied on an extremely shallow understanding
                      of how *politics* has worked, i.e., the actual
                      systems in use in highly visible public
                      elections, rather than on fundamental election
                      theory, and more deeply than that, how
                      *individuals* make choices, and then groups of
                      individuals, starting with two and increasing.
                      Without an exploration of these *fundamentals*,
                      about which a great deal is known, the political
                      science perspective on voting systems is
                      unrooted, without a foundation, and becomes mere
                      opinion conditioned by views of history, and thus quite limited.

                      Single-winner deterministic voting systems have
                      flaws, all of them, and these are well-known. For
                      that reason, direct democratic institutions avoid
                      making *any decision* without a number of
                      conditions appearing. It is likely that all of us
                      have experienced this: we set some standard or
                      procedure for making a personal decision of
                      complexity. Perhaps we decide to assign weights
                      to considerations, and then sum them. And we come
                      up with an answer that we don't like. Our
                      internal majority criterion *fails*, and it fails
                      *instinctively*. It knows something that we don't
                      know, quite possibly. That is, we followed a
                      rational process. From the single-poll, yes/no,
                      overall judgment process, faced with the "range
                      vote" of our analysis, we now know that something
                      is missing, and we don't necessarily know what it is.

                      We are designed for two kinds of decision-making:
                      fast, run by the amygdala. This process I
                      conceptualized as an *immediate* range vote, it's
                      the strength of impulses in various directions
                      that are compared and the body is turned over,
                      immediately, for action, to the overall strongest impulse.

                      What we may miss is that the cerebral cortex is
                      *also* submittted to that amygdala choice. It
                      proceeds to generate rationalizations and plans,
                      which can feed the other decision-making process.

                      That other process is reflection and analysis and
                      consideration. When we are not awake to the
                      process, this other process, we call thinking and
                      which may then be supported by writing and other
                      exercises of language, seems like an attempt to
                      discover the truth, the best decision, etc.
                      However, if there has been an instinctive
                      response, and, again, this is well known, these
                      analytical processes serve that initial
                      instinctive response, they rationalize it, they
                      place weight on evidence supporting it, and they
                      discount or ignore evidence and ideas that seem to contradict it.

                      When there is a conflict between range analysis
                      and instinctive response, that's a sign that
                      something is unrecognized and unresolved. Hence
                      in standard deliberative process, a binding vote
                      does not take place until there is a
                      *supermajority* in support of that. Range
                      analysis is not used in binding votes, and that
                      is completely proper. But this is within a
                      context of full deliberative process. What is
                      being considered by Felsenthal has been
                      restricted to single-winner deterministic
                      process, with one exception, and he does not
                      recognize that it's an exception, because he has
                      so limited and confined it, it appears, that he
                      doesn't recognize the possibilities.

                      His kind of thinking is normal for those bound by
                      the past. It simply doesn't recognize the world
                      outside of expectation and history. He compounds
                      this limitation by not having considered the
                      great volume of voting systems theory, including
                      the work of Warren Smith, but certainly not
                      limited to that. One wonders what he thinks about
                      Arrow's recent CES interview, where Dr. Arrow
                      fully confirmed the possibilities of voting on
                      ordered categories rather than pure individual candidate preference order.

                      So, returning to the problem set by Felsenthal,
                      he's totally missed the *foundational* criterion
                      for judging voting systems, the maximization of
                      social utility. Because of this, he thinks that
                      violation of the Majority Criterion by Range
                      Voting and certain other systems is fatal,
                      instead of understanding the *nature* of the
                      violation, and without understanding that a truly
                      optimal voting system, with fully sincere votes,
                      *must* violate that criterion, on occasion, if
                      the electorate is inadequately *informed.* And a
                      poll is a method by which the electorate informs itself.

                      >I define a ‘voting paradox’ as an undesirable
                      >outcome that a voting procedure may produce
                      >and which may be regarded at first glance, at
                      >least by some people, as surprising or as
                      >counter-intuitive.
                      >I distinguish between two types of voting
                      >paradoxes associated with a given voting
                      >procedure:
                      >a) ‘Simple’ or ‘Straightforward’ paradoxes:
                      >These are paradoxes where the relevant data
                      >leads to a ‘surprising’ and arguably undesirable
                      >outcome. (The relevant data include, inter
                      >alia, the number of voters, the number of
                      >candidates, the number of candidates that must be
                      >elected, the preference ordering of every voter
                      >among the competing candidates, the amount
                      >of information voters have regarding all other
                      >voters’ preference orderings, the order in which
                      >voters cast their votes if it is not
                      >simultaneous, the order in which candidates are voted upon
                      >if candidates are not voted upon simultaneously,
                      >whether voting is open or secret, the manner
                      >in which ties are to be broken).
                      >b) ‘Conditional’ paradoxes: These are paradoxes
                      >where changing one relevant datum while
                      >holding constant all other relevant data leads
                      >to a ‘surprising’ and arguably undesirable
                      >outcome.

                      Notice that preference strength is not listed in
                      "relevant data." He's using the classic Arrovian
                      social-ordering-input of preference order, which
                      deliberately sets aside any expression of
                      preference strength. Thus his standards make
                      equal a vast gulf in expressed utility, and a
                      minimal or even arbitrary and forced minimal
                      preference strength. Thus this approach is
                      guaranteed to fail at the fundamental purpose of
                      voting, which can be variously stated.

                      I'd expect a true political scientist to
                      understand that fundamental purpose, and the
                      common and long-standing philosophy of it. But,
                      again and again, I see a total neglect of the
                      fundamental issues. Why do we vote? Why not just
                      do what the King says to do? or the Supreme
                      Council? (But does the Supreme Council vote? Why?)

                      The Montesqueian separation of judgment and
                      execution is not recognized, i.e, the separation
                      between evaluation and judment and *active
                      decision* is ignored. *That separation exists in
                      us as individuals, when we are sane, and it
                      exists in every functional social structure.*
                      Never confuse the process of rationalization and
                      analysis with the process of making a choice.
                      When we have time, we may engage in the former,
                      again if we are sane, but what that does, at
                      best, is raise considerations for our
                      *instinctive choice mechanism* to operate on. The
                      actual choice is not conditioned, in itself.

                      Underneath all choices is a Yes/No process on a
                      single pair, but that pair can be complex, i.e.,
                      X vs (not-X). A process that limits the choice to
                      two defined alternatives, with an a-priori
                      rejection of *all other possible choices,* is a
                      choice pre-defined to ignore the optimal. How we
                      ended up with such systems in public elections is
                      a long story, but I can guarantee this: it was
                      not based on a thorough review of all options,
                      and such systems continue because of inertia.
                      Inertia is not necessarily a bad thing, by the way! But it's limiting, right?

                      Felsenthal considers only two systems that are
                      not deterministic on the first ballot. I'll address each separately.

                      >2. Plurality with a Runoff: Under the usual
                      >version of this procedure up to two voting rounds
                      >are conducted. In the first round each voter
                      >casts one vote for a single candidate. In order to
                      >win in the first round a candidate must obtain a
                      >plurality and a minimal percentage of the
                      >votes (usually at least 40%). If no candidate is
                      >declared the winner in the first round then a
                      >second round is conducted. In this round only
                      >the two candidates who obtained the highest
                      >number of votes in the first round participate,
                      >and the one who obtains the majority of votes
                      >wins. This too is a very common procedure for
                      >electing a single candidate and is used, inter
                      >alia, for electing the President of France.

                      Felsenthal misses the obvious. Plurality with
                      elimination of all but the top two is obviously
                      flawed, but Felsenthal misses that this is not
                      the standard deliberative process. He is here
                      only thinking about systems in actual usage in
                      public elections, but with the next example of a
                      multiple-option, multiple-poll system, he's
                      clearly including deliberative process. As a
                      result, he misses, for example, what was just
                      proposed in Arizona: a two-round system, approval
                      in the first round, with top-two presentation in
                      a general Plurality election, and write-ins still
                      allowed, *so the system is not guaranteed to find
                      a majority in two rounds.* Extensions and
                      modifications to that are obvious. For example,
                      the general election could also be Approval.
                      Felsenthal does not apparently understand that
                      any ideal voting system, to face real conditions
                      in public elections, and satisfy the criteria
                      that he considers optimal while also maximizing
                      overall voter satisfaction, *must* be possibly
                      non-deterministic on the first round.

                      Now, Plurality works with a two-party system,
                      where the "primary rounds" are for each party.
                      When there are three or more parties, more sophistication is needed.

                      >4. Successive Elimination (Farquharson, 1969):
                      >This procedure is common in parliaments
                      >when voting on alternative versions of bills.
                      >According to this procedure voting is conducted
                      >in a series of rounds. In each round two
                      >alternatives compete; the one obtaining fewer votes is
                      >eliminated and the other competes in the next
                      >round against one of the alternatives which has
                      >not yet been eliminated. The alternative winning
                      >in the last round is the ultimate winner.

                      Sometimes a parliament or assembly may have a set
                      procedure for choosing from among multiple
                      options. The generic process, however, is the
                      amendment process. If a motion has been seconded
                      and accepted for debate (generally it takes a
                      two-thirds majority to prevent debate), the
                      motion is then open for amendment, and each
                      motion for amendment is then a separate question
                      with a Yes/No vote. To pass and be accepted, a
                      motion must gain a majority. The "two
                      alternatives" in this case are Yes or No on the
                      motion for amendment. Debate on each amendment
                      and then on the main motion, as amended, if any
                      amendments pass, is closed with a motion for the
                      Previous Question, which requires a two-thirds
                      vote. Then there is, assuming debate is closed, a
                      vote on the motion which will pass or fail. It is
                      also possible to reconsider a vote, and
                      Reconsideration, in my memory of the rules,
                      requires a motion by someone who voted on the
                      prevailing side, must be seconded, and must pass
                      by a simple majority, in which case the main
                      motion is back on the floor for review de novo.

                      Felsenthal looks at one particular process,
                      ignoring the general process. He is interested in
                      multiple simultaneous questions, which the rules
                      of order generally attempt to avoid, for all the
                      obvious reasons. For efficiency, it's done.
                      (Theoretically, an office can be filled by a
                      simple motion appointing a person to fill the
                      office, majority required. That's amendable by a
                      majority. So this process is Condorcet-compliant,
                      we might notice. If a cycle arises, a theoretical
                      possibility, someone might object to the
                      renomination of someone previously nominated, and
                      ultimately the assembly would decide how to
                      handle it. The assembly is *always* the sovereign
                      in democratic process. Assemblies *routinely*
                      follow the rules, which were designed for routine
                      efficiency, but the majority is not *bound* by
                      the rules, it always has the power to interpret them and to make exceptions.)

                      Robert's Rules has an extensive discussion of
                      "Filling in the Blanks." But what Rosenthal is
                      describing is not Filling in the Blanks, it may
                      be a specific process used in, say, a Parliament
                      where he knows the procedure. He's missed
                      something *crucial,* at least if that procedure
                      is what I'd expect from generic parliamentary
                      procedure. The "last round" in the procedure
                      described does not determine the outcome. It
                      determines a *nomination* for a Yes/No vote on
                      the measure. That's all. If the result is
                      majority No -- or if a ballot is used and a
                      majority vote No *or mark the ballot in any way
                      other than by leaving it blank,* the motion
                      fails, the whole process is moot. A "Plurality
                      result" is not accepted *unless a majority
                      approve it.* There is, in this case, no winner,
                      and no prejudice. (Other than certain rules that
                      make it more difficult, perhaps, to re-introduce
                      a failed motion in the same session.)

                      Essentially, Felsenthal totally misses how real
                      deliberative process works, and real deliberative
                      process incorporates elements of many different
                      voting systems, is Condorcet-compliant.

                      Further, Felsenthal misses, as well, that the
                      electorate shifts in multiple rounds. He makes
                      the assumption of a constant electorate, whereas,
                      in real repeated elections, more strongly
                      motivated voters (i.e., high preference strength
                      over the candidate set) are more likely to vote,
                      and weakly motivated voters to not vote. This
                      then shoves results toward a social utility
                      maximization. (If we assume that strong
                      motivation represents a high difference in
                      perceived utility, and if we want to use
                      perceived and expressed utility as a way to
                      maximize true social utility, i.e, as a proxy for
                      it. And I don't see a better alternative, ever.
                      We will generally assume that someone not
                      motivated to vote has no strong opinion. That
                      assumption, of course, fails, isn't totally
                      accurate, because there are many people who vote
                      in the absence of strong preference, encouraged
                      by a constant barrage of propaganda about voting,
                      that voting itself is a positive social value.
                      I'm not saying that this is wrong, and, properly,
                      it would be accompanied by a strong social value
                      to actually becoming informed! Which is a lot of
                      work, to be done properly, and people have other
                      values in life. In a sense, in the U.S., anyway,
                      most of us leave the political process to
                      self-defined *experts on politics.* It's just
                      what's so, I'm not claiming "right" or "wrong."

                      In looking at Successive Elimination in his chart
                      of paradoxes, he has it show the "Condorect
                      winner {cyclical majorities) paradox. On this paradox, he notes:

                      >1. The Condorcet (or voting) paradox (Condorcet,
                      >1785; Black, 1958): Given that the
                      >preference ordering of every voter among the
                      >competing candidates is transitive, the
                      >(amalgamated) preference ordering of the
                      >majority of voters among the competing candidates
                      >may nevertheless be intransitive when the
                      >various majorities are composed of different
                      >persons. All known voting procedures suffer from this paradox.

                      Notice the given. He claims that all "voting
                      procedures suffer from this paradox." There is no
                      paradox. First of all, there are voting proceures
                      not based on a preference ordering. We may derive
                      a preference ordering of *some candidates* from
                      Range votes, but preference ordering, with
                      Successive Elimination, for example, is not
                      expressed in the repeated polls, it is, rather,
                      something that might be *inferred* from the
                      polls, from a series of pairwise preferences. If
                      the goal of the process is to produce a result
                      satisfactory to a majority, the existence of what
                      is essentially an irrelevant fact about the
                      amalgamation of preference orders, is irrelevant
                      to the performance of the system. He has not seen
                      that Successive Elimination, if properly done --
                      i.e., as done by every deliberative body of which
                      I'm aware -- requires majority approval for a
                      winner. It might be as simple as a ratification
                      vote on a presumed election result, *but the
                      majority must approve of the election, explicitly.*

                      In the Robert's Rules procedure of Fill in the
                      Blanks, in some versions, each nomination must be
                      approved *as a nomination* by a majority. The
                      procedure is then selecting an optimal result
                      among majority-approved choices. (Hey, similar to
                      the Arizona bill, HR2518, except that it doesn't
                      specify "majority approval," it is top-two approval.)

                      What Felsenthal has done with Successive
                      Elimination is to assume underlying preference
                      order, first of all, when voters are only
                      considering a pair at a time, in the procedure he
                      describes. Secondly, he assumes that the process
                      is deterministic, when, in real practice, it can
                      fail to make a choice. All options can be rejected.

                      Range voting does *not* suffer from the Condorcet
                      paradox, what suffers is a voting systems
                      theorist who tries to analyze Range Voting from a
                      preference order perspective, considering
                      individual preference orders as transitive, first
                      of all -- we actually do have internal Condorcet
                      paradoxes -- but Range Voting generates an
                      ordering that simply does not consider the
                      pairwise comparisons, not directly. Range
                      satisfies a fundamental voting system
                      consideration that is not even listed by
                      Felsenthal, social utility maximization.

                      What Range Voting *actually suffers from*, if we
                      want to call it "suffering," is an ability to
                      choose a winner who apparently maximizes social
                      utility, but who is different from the first
                      preference of a majority. Felsenthal is
                      blissfully unaware, it seems, of the core issue.
                      And that is likely because his interest is really
                      not democratic process, but *political process.*
                      So he confines himself to cases within the realm
                      of his field of study, even though he tossed in
                      Successive Elimination. Because of his
                      background, he misses a fundamental aspect of
                      Successive Elimination, the requirement of
                      majority approval of a result. And, again, with
                      his interest in *political process*, as a
                      practical matter, he may reject successive
                      polling, allowing only single runoff systems
                      because he must, because they are in actual
                      usage. The real voting system used with far
                      higher frequency by deliberative bodies, electing
                      officers, is ignored as impractical.

                      And, then, hybrid systems that might simulate,
                      with reasonable accuracy, that standard
                      deliberative process, are also not on his radar.
                      Bucklin, for example, simulates a series of
                      approval polls with lowered approval cutoff,
                      which is pretty much how I'd expect a series of
                      approval polls to work in a deliberative body.
                      First poll, people would simply vote for their
                      favorite, at least most of them. But with
                      successive polls, if no result is found, people
                      would add compromise votes, at a position in the
                      process that represents, fairly, their
                      *preference strength.* Again, a total lack of
                      understanding of the significance of preference
                      strength runs through Ffelsenthal's work.

                      And then, of course, it's far from his thinking
                      that, say, a runoff system, with sophisticated
                      analysis of an original poll generating an
                      optimal set of candidates for a runoff or general
                      election, and with a sophisticated voting system
                      in the runoff/general election, could be
                      compliant with many of the intuitive criteria,
                      and only fail where social utility requires it.

                      I personally consider that majority approval is
                      required for a result, and particularly that a
                      result should not be determined without majority
                      approval, except under very narrow circumstances.
                      I.e., in a final election, where the jurisdiction
                      has decided, properly, that a plurality result is
                      necessary, even if there is no majority, I could
                      see, in a runoff, the Range winner being
                      declared, which would violate, technically, the
                      Majority Criterion, but this would be a *fully
                      informed* majority. It is *extremely* unlikely
                      that if this happens, there is strong preference
                      on the part of the majority, and it has, by its
                      votes, explicitly declined to express strong
                      preference, allowing other voters with strong
                      preference to determine the result.

                      Felsenthal has no concept of such tolerance on
                      the part of the majority, when, in real, healthy
                      deliberative bodies, *it's routine.*

                      The paper is full of such shallow analyses.

                      >– Minimization of the temptation to vote
                      >insincerely: Although all voting procedures are
                      >vulnerable to manipulation, i.e., to the
                      >phenomenon where some voters may benefit if they
                      >vote insincerely, some voting procedures (e.g.,
                      >Borda’s count, Range voting) are susceptible
                      >to this considerably more than others.

                      Generally, he misses that Borda's count and Range
                      voting are the same system, with only some
                      restrictions placed on voters in Borda's count.
                      He doesn not specifically define "insincere."
                      There are two basic kinds of "insincerity," and
                      only one is truly real, i.e, actually insincere.
                      True insincerity consists of reversing
                      preference. I.e, one *actually* prefers A over B,
                      but ranks B over A in a preference order system,
                      or rates B over A in a rating system like Range. Range *never* rewards this.

                      (There is a possible exception in a
                      multiple-round Range system, a far more complex
                      issue of an allegedly rational turkey-raising
                      strategy. It does not apply to Range as a
                      single-round deterministic system -- or as the
                      final round in a runoff system with more than two candidates.)

                      When I've seen theorists like Felsenthal
                      criticize Range as "vulnerable to temptation to
                      vote insincerely," i.e., as encouraging an
                      allegedly insincere vote, it has always been
                      based on a contradictory assumptions: that the
                      voter *really* has some sort of absolute
                      preference strength, but they *insincerely*
                      exaggerate it. What has been totally missed is
                      that there is no specific sincere vote in a Range
                      system. There might be a vote that is what one
                      would cast with no knowledge of how other voters
                      will vote, and we might think that sincere, but
                      it also might totally suppress all preferences
                      except to divide the candidates into two classes.
                      The issue of translation of some internal
                      "sincere utilities" to a Range vote is one that
                      these theorists have never examined. We do *not*
                      use, in our routine decision-making processes,
                      absolute utilities, we use relative or
                      conditional ones. We consider what possibilities
                      are "realistic," and we would not waste voting
                      power on irrelevant choices. We actually use, and
                      would likely vote in Range Voting, with maturity,
                      von Neumann-Morgenstern utilities, not absolute
                      utilities. We don't need the fancy name, we already know how to do this.

                      The "expert" is telling the voter what is sincere
                      and what is not. If a voter actually has a very
                      weak preference, they have *no motive* to
                      "exaggerate," and would be likely to rate two
                      candidates with such a weak preference between
                      them as the same rating or close. Or,
                      alternatively, they might divide the candidates
                      into two classes, approved and not approved, and
                      the preference strength between a particular pair
                      of candidates, one from each set, might be weak.
                      But all these votes would be sincere.

                      And that raises the possibility of an individual
                      condorcet paradox. A voter may, internally, have
                      intransitive preferences. And that would become
                      clear in the voting process. Ultimately, the
                      voter will make a choice. It's hard to imagine
                      that such a condorcet cycle would arise with
                      strong preferences, and that it would not be
                      resolvable with a meta-consideration. If the
                      preferences are weak, which is pretty much what
                      I'd expect, then the voter can just lump these
                      candidates in the same rating, totally avoiding
                      the paradox by not choosing between them. And if
                      that seems wrong to the voter, they will then
                      pull out the standout(s). We go through choice
                      processes like this *routinely*, but because
                      Felsenthal has not examined the foundations of
                      choice, he's clueless about this.

                      Now, to Warren's critique of Felsenthal:

                      >Some kinds of paradoxes, Felsenthal considers
                      >severe; other kinds bother him less.

                      It's remarkable that he considers a "paradox"
                      "severe" when violation is actually *necessary*
                      for optimal voting system performance. That's
                      because his concept of severity is not related to
                      real voting system performance, and he obviously
                      cares nothing about that. It's about compliance
                      with *his ideas* about what's necessary, his own
                      set of prejudices and knee-jerk assumptions.

                      Some of that may be based on some vague
                      understanding that has a realistic basis. There
                      *is* a problem with violation of the Majority
                      Criterion, when a majority has expressly
                      disapproved of a result, if that's clear.
                      However, by not understanding the matter of
                      preference strength, he's shut himself out of
                      understanding how to resolve the problem.

                      It goes back to the very nature of voting. Warren
                      has pointed out that the Spartans had a voting
                      system, I'll call it the Shout. In a meeting,
                      deciding upon options, the supporters of an
                      option would shout, and the loudest shout was the
                      social choice. This is obviously informal range
                      voting. A minority loudly shouting will be louder
                      than a majority weakly shouting. Show of hands
                      incorporates no preference strength information,
                      but preference strength is crucial for maximizing
                      true social utility and, this is fundamental,
                      avoiding conflict over the choice.

                      Consider a meeting among peers, and these are all
                      warriors, all prepared to fight, and generally
                      strong, except perhaps for some older ones.

                      So options A and B are considered. A majority
                      prefer A, but weakly. They will shout, perhaps,
                      but not with determination and power, because ...
                      they have a weak preference, and they are not
                      about to actually fight for it. They are just expressing it.

                      The minority who prefer B, I will assume, have
                      strong preference. Deny them that choice, they will actually fight.

                      Now, in that fight, who would win? I can tell you
                      where my money would be: on the minority with
                      high preference strength. Other factors being
                      equal, the sane social choice is on ... the
                      loudest shout. That takes adequate numbers. An
                      individual shouting loudly will ... be surrounded
                      by others who will restrain him, and especially
                      his friends. "Simmer down, now, we don't want to
                      see you dead because we love you. Let go of this
                      one! We'll be with you when it's important!"

                      My sense is that majority became the general
                      standard because, in a society of peers, other
                      things being equal, the majority would win a war
                      fought over the decision. And we all benefit by avoiding wars, right?

                      However, that assumption can fail, and there are
                      historical examples where the majority made
                      choices that they rued, because the minority was
                      highly determined and revolted, and kicked the
                      majority hard where it hurts. Never ignore a
                      determined majority, at least give them access to the decision-making process.

                      So, while there is a serious problem with
                      violating the majority criterion, the social
                      utility maximizing criterion is stronger, where
                      it is necessary to make a choice between
                      violations. And the majority, I will assume, is
                      wise enough to understand when to insist and when
                      to let go and allow a minority its way. If not,
                      in the long run, the society is unstable and dangerous.

                      So I resolve the issue with runoffs, or with an
                      overall system that sets up a general election
                      with an optimal candidate set, and that empowers
                      the majority to make a wise choice.

                      Empowerment of the majority is fundamental to
                      democracy, as is allowing full and free
                      expression of opinion. I find it remarkable that
                      some voting system theorist believe that
                      elections are improved by confining voter choice.
                      Thus Saari, for example, thinks that Borda is
                      superior, whereas Borda *is* range voting with
                      two restrictions: no equal ranking/rating, and no
                      empty ranks. Thus, he must believe, restricting
                      the choices voters make will improve results, by
                      preventing voters from making what Saari considers bad choices.

                      In other words, to be blunt, he's an arrogant ***.

                      Having been an arrogant *** myself, having a
                      constant predeliction to arrogance, being *smart
                      as hell*, I'd love to have a conversation with Saari about this.

                      >So, e.g, according to his chart, Borda and the three mentioned Condorcet
                      >methods are the best, and both are superior to range voting. Indeed,
                      >according to his chart even plain plurality
                      >voting is superior to range voting.

                      Sure. Satisfies the Majority Criterion, one of
                      his Gotta-Have-Its. He simply does not understand the issues.

                      >Sigh.

                      Indeed. Warren, I suspect that you think of
                      yourself, sometimes, as a voice crying in the
                      wilderness. If you'd ever like to get out of that
                      trap, talk with me. As has been the case with
                      myself, one might need to be willing to give up a
                      "settled identity," in favor of something far
                      beyond it as to the world of possibilities. Up to you.

                      >So now, let us examine the details.
                      >The three "severe" violations range voting suffers are
                      > 1.Absolute majority
                      > 2.Condorcet loser
                      > 3.Absolute loser
                      >all of which are simultaneously
                      >demonstrated by the following example:
                      >
                      > 51 voters: Hitler=5, Truman=4.
                      > 49 voters: Truman=9, Hitler=0.

                      I love the example. But was this example, or
                      yours? Yours has been designed to make his
                      criteria look ridiculous. It's also totally
                      ridiculous. You've used the name of a highly
                      polarizing demogue as your example of a poor
                      majority winner, yet the voting shows the Other
                      Guy as having polarized supporters. What is wrong with this picture?

                      But, yes, as a demonstration of a possible
                      problem with the majority criterion, it's
                      classic. And it really doesn't matter what those
                      names are, they merely create knee-jerk noise.

                      Rather, we have a minority with very strong
                      preference, and a majority with maximally weak
                      preference. Okay, suppose that Hitler wins the
                      election. What will those Truman voters do? If
                      their votes were sincere, they are out in the
                      streets, burning cars would be the least of it.
                      And the Hitler supporters would be staying home,
                      since, obviously, they don't have any strong
                      preference. And they might be, ah, changing their
                      minds. They might put Truman signs on their door,
                      once they realize the situation. They might be
                      cheering with their neighbors. *Nobody* would be
                      supporting Hitler, he'd never take office. He'd
                      be lucky to get out of the situation alive.

                      All the while believing that he won the election,
                      fair and square. If he believes that strongly
                      enough, he might die for it. And so might the few
                      strong supporters. (Where were they in that poll?
                      This is totally unrealistic, just get that.)

                      >Range voting elects Truman, which according to Felsenthal
                      >is a "severe paradox" because it evidently is his view that
                      >Hitler should have been elected.

                      He's got a knee-jerk response that violating the
                      majority criterion is severe. That's all. He
                      might actually be more reasonable than it
                      appears. But you have the book, I don't. You
                      cannot actually claim that it's his view that
                      Hitler should be elected, unless he actually said that.

                      Warren, it's highly offensive, it upsets people,
                      when you put words into their mouth, based on
                      *your* logic from how you understand what they
                      said. In the Socratic method, you can make
                      logical deductions from expressed opinions, but do remember two things:

                      1. They condemned Socrates to death.
                      2. It's far more effective to *ask* about those
                      conclusions than to assert them as the view of the other person.

                      I.e., "This election presents a voting pattern
                      where it is apparent that the majority only
                      weakly supports its first preference, and a very
                      large minority strongly opposes that weak
                      majority preference. With these votes, who should win this election?

                      Then you let *them* stick their foot in their
                      mouth, if that's what they choose. My guess is
                      that Felsenthal is a bit too smart to go for the
                      trap, but it could be worth pursuing the
                      conversation. From my assessment, and somewhat
                      from your report of him, he would prefer to avoid
                      the conversation entirely, sensing that there is
                      no way for him to "win." But I'd love to be wrong
                      about that. I'm dealing with some very
                      strong-willed scientists, survivors, able to
                      stand against widespread rejection, sticking to
                      clear evidence and logical inference, and some of
                      them are utterly intolerant of disagreement, but
                      others will actually listen and modify their
                      views, or at least their expression.

                      There *are* people who are willing to learn, even
                      if they are already world-class experts.

                      >Of course, a different reading of the situation would be
                      >that all three of those logical criteria actually should
                      >not be demanded! Felsenthal was simply wrong to label
                      >the violation of those criteria a "paradox" much less
                      >brand it "severe." But of course Felsenthal does not
                      >even mention that. It does not even occur to him.

                      Right. He's floating. He has no grounded basis
                      for his opinions. If he wants to reject the
                      Social Utility criterion, fine, but his opinions
                      show no awareness of the issue at all.

                      (It's classically rejected because there is
                      supposedly no way to obtain absolute
                      commensurable utilities. Yet Range Voting does
                      not ask for those. It simply asks for weighted
                      votes. I typically explain Approval as Plurality
                      with the voters being allowed to vote for more
                      than one, and Range as allowing voters to cast
                      fractional votes. What Range does is to *allow*
                      the expression of what might be relative
                      utilities, still following the standard
                      assumption of the equality of voters. Voters are
                      *allowed* to cast weak votes, that's all. As in
                      Approval Voting, voters may vote Plurality style,
                      so too with Range Voting, they may vote Approval
                      or Plurality style. It is their choice. The
                      system has given them additional flexibility, and
                      standard sum-of-votes does fairly use that
                      information, and it is especially fair if the
                      voters know how the information will be used,
                      i.e., that these are *votes*, not "sentiments.")

                      (Since Felsenthal has no concept of the
                      importance of preference strength, he, as well,
                      seems totally unaware of Bayesian Regret and the
                      usage of voting system simulations. He only has
                      raw abstract reasoning, unrooted in actual voting
                      system behavior, except for a pile of public
                      election results using existing systems -- all
                      known to be defective in major ways.)

                      >
                      >Another thing to consider is the fact that all three
                      >of these violations come with asterisks:
                      > 1.If an absolute majority wants X, they can
                      > force X's election with range voting.
                      > 2.In the real world where many voters are
                      > strategic, there is good reason to suspect,
                      > backed up by computer simulations, range will
                      > elect condorcet winners in practice more often
                      > than condorcet methods do(!), see
                      > http://rangevoting.org/AppCW.html
                      > 3.If an absolute majority does not want X, they can force X to lose
                      >with range voting.

                      None of this matters to him. He's blinding by the
                      criteria that he considers important, he's not
                      able to see even obvious contrary evidence or possibilities.

                      This matter of the power of the majority is
                      *crucial* to voting systems. I'll say, Warren,
                      that you have missed it. You have taken the
                      position that the Range winner should prevail.
                      But this neglects the reasonable possibility that
                      this is just an appearance, caused by voter
                      ignorance. Rather, when there is a conflict
                      between a Range winner and a majority preference,
                      there is an obvious solution: a runoff. In
                      studying runoffs, you may have assumed that the
                      electorate would be the same, and with the same
                      preferences. That is *extremely* unlikely. I've
                      written about this at length, that in a standard
                      runoff election, as an off-general election, the
                      Range winner in a primary, facing a different
                      candidate who was the majority preference) is
                      actually the leader and will probably win, and a
                      majority preference, sitting on his or her
                      laurels as that, is quite likely to lose. Until
                      your simulations have shown that, I'll likely
                      suspect that you haven't got the conditions
                      right. Turnout and the effect of additional
                      information must be considered. The simulation
                      will need to assume a larger electorate, much of
                      which does not vote unless they see a reason. The
                      supporters of a minority candidate may have
                      stayed away from the original election, thinking
                      their choice did not have a chance. Maybe.

                      The reverse may be true: the supporters of the
                      majority preference may have stayed away,
                      thinking their choice was assured. Some runoffs
                      have low turnout, some have high. It's important to understand why!

                      >In view of these 3 asterisks, calling these "severe" was ludicrous.
                      >But Felsenthal does not even mention the asterisks.

                      Right. He's utterly unsophisticated.

                      >The 4 non-severe paradoxes Range Voting suffers from (says Felsenthal) are
                      > 1.Condorcet "intransitive" cycle can exist.

                      Well, he says that every system is "vulnerable"
                      to this. This, however, doesn't affect Range
                      voting at all. So what if it exists? Range
                      doesn't use pairwise preference *at all*. We can
                      derive pairwise preference information from Range
                      votes, to a limited extent. Felsenthal seems to
                      have no distinction between the actual voting
                      system and what actually happens, and his own analysis of what might happen.

                      That's a sign of the well-known HUB syndrome. HU
                      stands for "Head Up," I'll leave the B undefined, i.e., to the imagination.

                      > 2.Strategic voting: there can be situations where voter is
                      >better off lying in her vote than honest voting.

                      He's dead wrong here. First of all, did he say
                      "lying"? If he does, that's nuts. Warren, if he
                      didn't say "lying," then it's unfair for you to
                      put that word in his mouth. What did he actually say?

                      If he's like all the others, he meant something
                      like, "better off exaggerating," as this applies
                      to Range. Which then raises the question
                      "exaggerating compared to what?" With pure ranked
                      systems, equal ranking not allowed, the only
                      "insincere votes" possible actually misrepresent
                      preference order, which could be called "lying."

                      Actually, though, voting isn't lying because it
                      is not testimony. It's just tossing weights in
                      baskets. The voter may assign meaning to those
                      votes, or not, and the system may assign power to
                      them, i.e., effect. There is no lie, there are
                      only free choices, free in the sense that the
                      voter may choose based on the voter's own
                      consideration of the votes and the effects and
                      the voter's sensibilities and motives.

                      There is a *huge* amount of hot air that's been
                      emitted over what votes supposedly "mean." They
                      may *indicate* underlying preferences, or not. It
                      depends on conditions. Calling votes "lying" or
                      even "exaggerated" sets up a series of knee-jerk
                      associations, demaging our ability to see
                      clearly. After all, didn't they teach us in
                      kindergarten that lying is bad. And, Johnny, now,
                      Don't Exaggerate! Tell the Truth!

                      >(Felsenthal claims every method he charts suffers both problems 1 & 2.)
                      > 3.Condorcet winner can exist but not be elected by range voting.
                      > 4.Truncation -- it can be strategically superior for some voter
                      >to leave some candidate unscored.
                      >
                      >OK, let us reply to these.
                      >Re 1, the condorcet cycle problem is a matter of self-inconsistency.
                      >That is why it is called a "paradox."

                      The paradox arises because we have a naive
                      knee-jerk assumption that a transitive
                      amalgamated preference order can be derived
                      consistently from individual transitive
                      preference orders. I.e., it is easy to assume
                      that individual preference orders are transitive,
                      that if A is preferred to B, and B to C, then A
                      is preferred to C. It doesn't necessarily hold,
                      as I've explored a bit, but it's still reasonable
                      as an operating assumption. And a system may take
                      a preference order as input, regardless of whether it's *real* or not.

                      Condorcet described the problem long ago, which
                      is why it's called the Condorcet paradox, of
                      course. It is quite possible, and not terribly
                      uncommon, that a majority of voters might prefer
                      A to B, B to C, and C to A. Is that a "paradox"?
                      Only from the point of view of a naive
                      expectation. The "majority" is not the same set
                      of voters in each pairwise consideration. If it
                      were, the cycle would be impossible. I think the
                      naive expecation arises because we think of
                      "majority" as if it were a thing, a block of voters. It's not.

                      >Range voting is not inconsistent, i.e. it can NEVER say that A>B>C>A
                      >where ">" means "has greater mean score."

                      Right. Range analyses votes in a very different
                      manner than do ranked systems. The scores
                      generated are numbers, being sums of real numbers
                      in a defined range. Range utterly side-steps the
                      Condorcet paradox, it doesn't care about it *at
                      all.* However, for the same reason, it can show
                      Majority Criterion failure. But that's totally
                      different, a different conversation. Felsenthal
                      is here showing his knee-jerk thinking. He's got
                      a paradox, he's declared that it's universal,
                      that all voting systems are "vulnerable" to it,
                      and, so, when it comes to Range, there is no
                      thinking to do. It's automatic. It's a voting
                      system, and *therefore*, in his mind, it is
                      "vulnerable." Does he show how, in the book?

                      > This differs
                      >from (say) Black's method, where Black can easily say that A
                      >would beat B if the other rankings (for C) were erased
                      >from all ballots. (And etc for B vs C, and C vs A.)
                      >
                      >Therefore, range voting does not suffer a "paradox" of this kind,
                      >contrary to Felsenthal.

                      Yup. Right on, Warren. He has confused pairwise
                      analysis with a voting system that doesn't use it.

                      Now, I'm proposing systems that do, in fact, use
                      pairwise analysis on a Range ballot, and with
                      that, a consideration of pairwise issues,
                      including the Condorcet paradox and the
                      possibility of cycles, is necessary. But that's a
                      hybrid system, used to determine runoff
                      candidates, to guarantee that a Condorcet winner
                      (or at least a member of a Condorcet cycle) makes it into a runoff.

                      The "paradox" only appears if we do pairwise
                      analysis. It's not a problem for the voting system itself.

                      >Of course, by redefining the
                      >word "paradox" to mean "I, Dan Felksenthal do not like it"
                      >that difficulty is avoided.

                      Indeed.

                      > However, using
                      >the dictionary definition, this is no paradox, and indeed t
                      >his is an advantage of range voting versus every other method
                      >on his chart except MJ. Felsenthal fails to comprehend that.

                      Yes.

                      >Also, although a consideration of voting systems using
                      >the "Felsenthal does not like it" subjective metric
                      >perhaps is also of some interest, the fact that Felsenthal
                      >prefers that Hitler win in above scenario, means his judgment is poor.

                      Again, does he say that? What does he actually
                      say, Warren. Allow me to suggest that, right now,
                      to empower you in conversations with others, you
                      give up the habit of putting words into the
                      mouths of others, no matter how "logical" it
                      might seem. You *really* don't like it when
                      others do that with you, take a hint from that.

                      Of course, if he actually said that, you can
                      quote him. I am *not* claiming you are wrong,
                      Warren. I just rather doubt that someone in his
                      position would actually say, where he might be
                      quoted, that Adolf Hitler should have won an
                      election against Harry Truman. AMIRITE?

                      >Re 2, Felsenthal LSE version QUOTE
                      >Ceteris paribus, a voter may obtain a preferred
                      >outcome if he votes strategically, i.e, not according to his true
                      >preferences. All known voting procedures suffer from this paradox.
                      >END QUOTE.
                      >
                      >Felsenthal is actually correct that under that definition
                      >range voting does suffer this problem,
                      >but gives no proof and I suspect based on his general level of
                      >ignorance and incompetence that he is incapable of producing a proof.

                      Actually, Warren, you have never recognized the
                      problem here, yourself. The difficulty in in the
                      definition of "true preferences." Generally, a
                      voter's preferences as will be expressed in Range
                      votes, are not fixed things and will include
                      choice based on a percenption of *how others will
                      vote,* which can be considered from two different perspectives.

                      1. The voter doesn't want to cast a weak vote
                      where it might matter to cast a strong one.
                      2. The voter wants to maximize *overall* social
                      utility, i.e., cares about others.

                      It was recognized long ago that a "problem" with
                      Approval Voting was that there was no absolute
                      category called "approval." Rather, we approve
                      *choices,* and we make our choices based on
                      *expectations.* I.e., standard approval strategy:
                      approve all candidates with better than expected
                      utility. "Expected utility" is a sense of what is possible in the election.

                      If I expect that A is likely to win the election,
                      and I prefer A, but B is close, it does not
                      matter if I have weak preference between A and B,
                      I am likely to vote for A and not for B (and,
                      possibly, for nobody else as well). I will not
                      vote for B based on someone telling me that I
                      "also approve" of B, even if I might actually be
                      pleased by the election of B, under other circumstances.

                      Running like a corrupted vein through a great
                      deal of Approval and Range discussion is the
                      concept that there are absolute approvals or
                      absolute ratings, that, if we are "sincere," we
                      will vote. In Approval, we set an approval
                      utility cutoff, *where we choose*, and that
                      choice is free. We will set it according to
                      expectations, and it is for us to decide what we
                      expect. We then approve all candidates above the
                      cutoff, and don't approve all candidates below
                      it. We can set that cutoff *anywhere*, and then
                      all votes consistent with it are sincere, and
                      there is no advantage to be gained by voting contrary to it.

                      The same general understanding applies to Range.
                      We may postulate that there are internal
                      transitive utilities, though that's probably
                      inaccurate. At any given time, we have an
                      internal range voting process, comparing how loud
                      the shouts are, so to speak. This, however, can
                      be senstitive to *sequence of consideration.* We
                      have some difficulty comparing multiple options
                      at a time, more than two. We can do it, under
                      some conditions, but we generally simplify the
                      problem to pairwise comparisons. And since we, in
                      a sense, are just a pile of neurons, we can run
                      into the same Condorcet paradox as preference
                      systems. However, we do have a capacity to pull
                      out a louder voice from a crowd. But this may change from moment to moment!

                      At some point the voter pulls the lever, so to
                      speak. Mostly, it's easier to actually do it than
                      to describe the process of choice!


                      >Re 3, see above "severe" issue #2, same response applies.
                      >
                      >Re 4, Felsenthal gives this example to "prove" his contention:
                      >
                      > CANDIDATE....V1....V2....V3....V4...V5...V6...V7...mean score
                      > X.........1.....1....1.....10....5....4....7.....4.143
                      > Y.........2.....2....2......3....8....5....8.....4.286
                      >
                      >In this situation (Felsenthal says) Y wins with range voting.
                      >Voter 4 (denoted V4) is unhappy about that. Voter 4 is better off
                      >NOT SCORING Y, because then Y would have won. Q.E.D.
                      > Unfortunately for Felsenthal, he's wrong. The rules
                      >of range voting he cites [namely rangevoting.org web page,
                      >and also the rules he himself describes in his section 3.2.2.5]
                      >say highest average wins.

                      Looks like he is using 1-10 range.

                      Yeah, bonehead error.

                      > If V4 leaves Y unscored, then X still wins,
                      >indeed by more margin. Felsenthal invents on the spot the new rule,
                      >which he had not stated when describing the rules of range voting, that
                      >"no score" counts the same as "0 score." Under that new rule,
                      >Y would win.

                      What I notice is that this is a very weird
                      election. Most voters are voting extremely
                      weakly. Voter 1, for example, really doesn't like
                      anybody. They may have believed that they should
                      rate both candidates. Everyone casts a weak vote.
                      The strongest motivated voter, V4, still casts a weak vote. Why?

                      Now, perhaps there are other candidates in the
                      election, and only the votes for X and Y are
                      shown. Perhaps they have the highest mean scores.
                      But there is nothing to explain this bizarre
                      voting pattern. Is the minimum score a 0 or a 1? Nobody voted zero.

                      Okay, so suppose the system is Range 10, all
                      right, and one votes 0 by not marking the
                      candidate's score. But that is a *vote*, it is
                      merely a default. It is not "not voting." Rather,
                      the voter chooses to score in the range 1-10 or
                      to score 0, by not marking the canddiate. Perhaps
                      these voters did vote 0, but for other candidates
                      not shown. That could make some sense.

                      To repeat the paradox he's looking at:

                      > 4.Truncation -- it can be strategically superior for some voter
                      >to leave some candidate unscored.

                      That's not possible in the system described. A
                      voter who votes for *any candidate* is "scoring"
                      all of them, by default at 0, as Felsenthal has
                      described the method. What is happening is that
                      the voter, by lowering their score for a
                      candidate, may cause that candidate to lose. Big
                      Whoop! Satisfies monotonicity criterion!

                      Warren, don't get stuck on the "average vote"
                      rule. It will just irritate people needlessly. My
                      position is that the world isn't ready for
                      "average vote," i.e., with blanks being
                      considered abstentions from rating a particular
                      candidate, rather than min rating (or some other
                      default, which would then have the problem that a
                      blank becomes a kind of vote *for* the
                      candidate.) In any practical implementation, in
                      the near future, use of average range, as
                      distinct from sum-of-votes range, which treats a
                      non-rating, generally, as if it were min rating,
                      is such a violation of standard democratic
                      practice that it's likely to torpedo the implementation.

                      An attempt has been made to address this with a
                      "quorum rule," but that runs into real or
                      perceived arbitratiness. Please, Warren, be
                      content with sum-of-votes range, it is far easier
                      to explain and justify, it is simply fractional
                      voting, and only a desire to enable dark horses
                      to win, which is *result oriented*, leads to a
                      consideration of the usage of average range. Such
                      a win, if it actually happens, could torpedo the
                      system forever, when voters *who can change the
                      system* realize what happened. Really, if you
                      will pay attention to the requirements of
                      majority rule, and what that means to democratic
                      tradition, as distinct from abstract theory,
                      you'll get it. We can have range optimization *and* majority *consent.*

                      Both. As was seen in Burlington, no amount of
                      rationalization can counter a majority that saw
                      their preference frustrated. A majority will not
                      object to its preference being submitted to a
                      runoff. But award the election to a minority,
                      without their explicit consent, you could be
                      sowing the whirlwind. I'd consider it a disaster
                      if a Range implementation finally comes, and it's
                      average range, because your arguments have been
                      accepted, and then the whole thing is dumped
                      because of some marginal sitation with a dark
                      horse and many voters not rating the candidate.

                      Or, in the extreme, a write-in. Only supporters
                      rate the candidate. This candidate *will* have
                      the highest average rating, it's almost
                      guaranteed. So what do you need for a quorum?
                      Hint: a majority would be minimal.

                      >Similarly, if I invented new rules on the spot
                      >when inventing paradoxical situations, I could
                      >make any voting method fall to any paradox I wanted.

                      Your argument is weak, here, it's essentially
                      quibbling over the rules for Range Voting. He
                      really assumed Range 10 with a 0 default. Big
                      deal. Even the meaning of "average vote" is
                      ambiguous. If only one person actually rates a
                      candidate, what is the average vote? The real
                      question is the *default vote,* how blanks are
                      treated. Felsenthal wanted to create a failure
                      from truncation, but the system he set up did not
                      allow truncation, actually, and he defined it
                      that way. That's an internal contradiction in his
                      argument. His approach was totally artificial,
                      and you are correct about this: he defined the
                      rules to make it seem he was showing an example
                      where not rating a canddiate improved the result,
                      therefore "voting" made it worse.

                      Repeating myself, the voter actually rated the
                      candidate both ways. It was not that one pattern rated and the other did not.

                      >OK, so let's tote up the score. Felsenthal makes 2 mistakes
                      >about range voting chart entries. He also
                      >misunderstands/misuses/redefines the
                      >meaning of "paradox" which in an article about paradoxes,
                      >is rather bad.

                      Indeed. The biggest error though, is *total
                      neglect of social choice theory.* There is a
                      confusion, by the way, between utility theory and
                      "utilitarianism," which is a philosophy that can
                      be quite different. Utilitarianism, as a
                      philosophy, sets certain norms, sets up some
                      ideals, about which we can easily disagree. But
                      the basic concept of utility is something else,
                      and how one can assess voting systems without any
                      method of assessing the quality of results is
                      totally beyond me. It becomes an exercise in pure appearance.

                      Suppose we have a sales force, and we want to
                      judge how effective salespeople are. We could
                      judge by "sales criteria," i.e., by whether or
                      not the sales people follow set rules of
                      behavior, what we expect a good sales person
                      would do. Perhaps we record the number of phone
                      calls they make, or listen in and grade them on certain ideals of performance.

                      Or, we could actually look at their sales
                      numbers. We could measure customer satisfaction through polling.

                      Nah, we want to use our "sales criteria." I'm the
                      Sales Manager and they better do what I like, or
                      they are history. Yes, Sir, ... What? You are
                      firing me? Why? I was trained in sales criteria,
                      all the right ways to do things, and all I've
                      done was make my sales team follow the rules! You
                      are making George the new sales manager? But he
                      was doing it all the wrong way, that's why I
                      demoted him. He had high numbers? How can you
                      compare mere performance numbers with following
                      the correct procedures? He was just lucky! Hello?
                      ... Damn! he hung up on me! Why is it that they
                      never understand me? I'm doing everything right!


                      >He also leaves many other paradoxes out of his
                      >considerations entirely, whicjh if they were included would change
                      >the picture.
                      >
                      >He states the following in his (LSE version) paper:
                      >QUOTE
                      >Among the 17 (deterministic) voting procedures
                      >analysed in this paper, the
                      >Condorcet-consistent procedures proposed by
                      >Copeland (1951) and by Kemeny (1969) seem to me to be the most
                      >desirable from a social-choice perspective for
                      >electing one out of several candidates.
                      >END QUOTE
                      >
                      >OK, that sentence reveals that Felsenthal is an idiot.

                      It appears that Felsenthal is not aware of the
                      problems with Kemeny as a practical system.
                      That's because Felsenthal is operating in a
                      vacuum. He's essentially talking off the top of
                      his head, with no broad experience.

                      >First of all, with Kemeny it is NP-hard to determine the winner, meaning
                      >in a large election we could not guarantee we
                      >would be able to determine the winner even with
                      >all computers in world running for years. That means Kemeny is
                      >not "the most desirable" it is "unacceptable for large elections."
                      >
                      >Second, with Copeland it is common for elections to be exact ties
                      >even with millions of voters. You can confirm this with
                      >a simple computer simulation, which obviously Felsenthal has never done.

                      On these, Warren, I'll assume you know what you
                      are talking about. You usually do, on matters
                      like this. I strongly suspect that the view that
                      Felsenthal is not familiar with voting system simulations is accurate.

                      Let's say that I could be shocked that anyone
                      presuming to be expert on voting systems would be
                      so totally ignorant of the state of the art, such
                      that he would only cob together some
                      shallowly-understood and poorly-digested material
                      and regurgitate it as something supposedly worth reading.

                      I'm going to quote again the thinking that produced the paper:

                      >I therefore thought it would be well to
                      >supplement that report by reminding social choice
                      >theorists, political scientists, as well as
                      >commentators, policymakers and interested laymen –
                      >especially in the UK and the US – of the main
                      >social-choice properties by which voting
                      >procedures for the election of one out of two or
                      >more candidates ought to be assessed,

                      He is attemting to inform experts on matters
                      where he is ignorant. Yes, he's an idiot, in
                      common parlance. He wouldn't survive a week on
                      the Election Methods mailing list. Below, I'll
                      mention what keeps him stuck in his idiocy.

                      >The job of a voting method, is to produce one winner. Kemeny and Copeland
                      >fail to perform that job. The question of what "paradoxes" that winner
                      >may suffer from, is secondary. First you have to DO your job, THEN we reckon
                      >whether it was a good job. If you never do your job, then you do not
                      >even get to enter the contest.

                      He has likely never engaged in discussions with
                      informed peers or experts, as to the issues
                      around voting systems. He is living in an ivory
                      tower, that protects him from that, my guess.

                      >I have told Felsenthal and Machover all this well before Felsenthal
                      >wrote this paper.
                      >He ignored me and wrote the paper anyway, never even mentioning these
                      >disqualifying problems with Kemeny and Copeland.
                      >So in addition to writing a paper containing nothing new, he also
                      >published mistakes.

                      Indeed. And, apparently, a book as well.

                      Now, Warren, I can easily conspire with you in
                      criticizing Felsenthal. However, that will do
                      *you* no good at all. What I see here is that you
                      failed to communicate with them. You can blame
                      them for that, but this will leave you with no
                      power. The real issue here, for you, is how to
                      communicate effectively. I can tell you the first
                      step: start by taking responsibility for the
                      *effect* your communications have, and giving up all the excuses.

                      What is your goal and purpose, Warren?
                    Your message has been successfully submitted and would be delivered to recipients shortly.