[EM] Eugene rejected 'IRV' option; vote-splitting in AV

[Craig Carey]

At 01.09.19 09:13 -0700 Wednesday, Forest Simmons wrote:
 >The following message may be of interest to EM list readers:
 >
 >Forest
 >
 >---------- Forwarded message ----------
 >Date: Tue, 18 Sep 2001 23:52:51 -0700
 >From: Fillard Rhyne <fillard at hevanet.com>
 >To: FairVoteOR <FairVoteOR at yahoogroups.com>
 >Subject: [FairVoteOR] Eugene
 >
 >The unofficial final results for the City of Eugene ballot measure,
 >from <http://www.co.lane.or.us/Elections/20010918.htm>, are 34.3%
 >in favor of preference voting and 65.6% against.
 >
 >Fillard


The CVD is still promoting the use of a truly defective method.

At least that seems to be the latest thinking of Rob Richie who
_still_ was pushing the IRV ios good enough to actually use opinion.



At 2001.05.28 20:06 +1200 Monday, Craig Carey wrote to instant-runoff-vote:

Knowledge held by the CVD:

 > > > We are particularly knowledgeable about: ... 3) the option of
 > > > instant runoff voting ... We believe it is essential to consider
 > > > voting system reforms that could reinvigorate American politics
 > >
 > >http://www.fairvote.org/about_us/index.html
...
A programmers version of the CVD method, the US state "invigorator's"
first fish mallet:

 > >
 > >    procedure IRV(BallotsVec) return set is
 > >    begin
 > >       if pub closing and no hand of cards
 > >          return (all meta-suites of poker bits;
 > >       else V2:=elim wee-est and back to table and shuffle hand and
 > >          return best IRV (BallotsVec);
 > >       end if;
 > >    end IRV-worlds-best-method;
 > >
 > >That is 4 SLOCS (semicolon-ed lines) long.
...

The Gang of 9 was running an anti-IRV campaign. The anti-IRV cartoons
are dated 2 Sept 2001 to 3 days ago (17 Sept), and they are here:

     http://www.thegangof9.com/past_cartoons.phtml

They were implying that IRV is too complex. It is nonmontonic. It may
have too many faces. I guess the CVD want to believe that the gang of
9 was being unfair by saying IRV was too complex.

Eugene is a rather small city region. It would be better if the CVD could
promote less dumb unfair methods than IRV and be less disliked presumably.

I saw this detail. The CVD mathematician and webmaster concedes the IRV
can be 'manipulated'. Manipulation was not well defined, but that can
be overlooked. Here is the text:

    "Multi-winner STV is even tougher to manipulate than one-winner STV."

  It is in paragraph no. 5 of the footnotes section, in
       http://members.aol.com/loringrbt/l_lor1.htm

Manipulated means buggy and fixable I suppose. I write to Mr Loring from
time to time, but I do not get a reply. The CVD probably does not want
any of its mathematicians talking to any other mathematician. Otherwise
communications over IRV could occur. Pity about failing to get the votes
needed in Eugene to introduce a wrong-winner-picking method that is
approx 4 SLOCs in primordial complexity.


-------------------------------------------------------

Here is another Hitler-Stalin example from Demorep where the middle
candidate of the 3 has the middle candidate right on the 1/3 quota. That
is able to lead to a total rejection of the argument by those that saw
the argument of casting 2 shadows, that is in
    http://www.ijs.co.nz/quota-13.htm
It is controversial to have 33 votes for B and 99 votes. That is done in
this example. The conclusion below (unless there is not one) is not
assisted by having controversy. Possibly no other subscribers here would
make that exact mistake.


At 01.09.19 03:43 -0400 Wednesday, DEMOREP1 at aol.com wrote:
 >Back to the past-
 >
 >34 ABC
 >33 BCA
 >32 CAB
 >
 >99
 >
 >Who, if anybody, has 50 YES votes (with 2 or more info bits per vote) ???
 >
...
That example does not use any "<" symbols which seems nice.
There does not seem to be a good argument there, Demorep.


--------------------------
Here are some examples from a private message I wrote.
I was attempting to show that IRV has a vote splitting problem just like
FPTP. The examples are not clearly showing that. This is one of the few
areas that IRV can seem better in. (A type of test it performs quite
badly in, is testing to see if it is "fair" (to individuals).)

 > -------------------------------------------
 >    AB    0
 >    AC   35   a = 35
 >    B    33   b = 33
 >    CA   14   c = 32
 >    CB   18
 >
 > Eliminate C:
 >
 >    A    35+14 = 49
 >    B    33+18 = 51
 >
 >    FPTP: A wins
 >    IFPP: A wins
 >      AV: B wins
 > -------------------------------------------
 >
 >
 > -------------------------------------------
 >    AB    0
 >    AC   39   a = 39
 >    B    31   b = 31
 >    CA   10   c = 30
 >    CB   20
 >
 > Eliminate C:
 >
 >    A    39+10 = 49
 >    B    31+20 = 51
 >
 >    FPTP: A wins
 >    IFPP: A wins
 >      AV: B wins
 > -------------------------------------------
 >

A was going to give votes for C and C was partially intending to
give votes to A. With IRV, winner B gets picked. That seems to
be the wrong winner if a legal idea of fairness is held.


Here is a simple version of the cascading of dominoes out of a
stairwell and up to the highest levels of democracy.
 >
 > -------------------------------------------
 >    AB    0
 >    AC   49   a = 49
 >    B    26   b = 26
 >    CB   25   c = 25
 >
 > Eliminate C:
 >
 >    A    49+0  = 49
 >    B    26+25 = 51
 >
 >    FPTP: A wins
 >    IFPP: A wins
 >      AV: B wins
 > -------------------------------------------
 >




PS. Mr Schulze wrote that IFPP is not monotonic, but the example was not
testing IFPP. Instead a not yet explicitly defined 4 candidate method
was tested. IFPP is defined to be monotonic (and to pass other tests).


G. A. Craig Carey