[EM] IRV is terribly unfair; The feline wrote off IRV

Craig Carey research at ijs.co.nz
Mon Mar 25 09:55:05 PST 2002

He. I am unsubscribing from the this list. I could reply to any replies,
and perhaps even re-subscribe.

At 2002\03\25 07:53 +1200 Monday, Craig Carey wrote:
 >Greetings Election Methods List members


My last message had partial re-lettering.

This is what the CVD person was replying to (I refer to the last message):

 >>· · · · · · · 1st · · 2nd
 >>· · · --------------------· · · ·change
 >>· · · ·A· · ·3335 · ·6666 · · · + 3331
 >>· · · ·BC · ·3333 · ·3333
 >>· · · ·C· · ·3332 · ·3334 · · · + · ·2
 >>· · · --------------------
 >>  AV Winner: · A · · · C
 >>· Total:· · 10000 · 13333

[The 'middle dots' have been added by me.]

The quoted words had not been altered. I got words including these:

    "(why would every B voter rank C second, but "

If the re-letter had been completely done then the text in the last message
would have been:

    "(why would every A voter rank B second, but "

It seemed to be preferable to get the example easy to remember while of
the exact same interpretation but 2nd property only now appears.

To compensate for the comments about Mr Richie I aim to withdraw from
opposing Mr Richie. My grounds were that mayors would be excessively
harmed by the CVD but I can't sustain this with the credibility of IRV's
fairness becoming unstuck and faltering so excessively. The EM List is
a wrong place to write this too. Maybe this message and the last could
be deleted by Mr Rob Lanphier say, before the 2003.


Whoops. There was a question asked and that was not attended to.

To search out for the truth that the CVD EXecutive Director seemed to
know, I asked a cat that I get to see at times, these two questions:

    "Why would every A voter rank B second?", and this too:
    "Why would every B voter rank C second?".

I got the intonation right so it would not have thought I was lying
to it or attempting to construct an evil plan for it, e.g. something
leading to its death or a loss of food only because it was that alone
that it was hoping for (vote negating).

The experiment ended and the cat did not reply.

I decided recently that the voters would be cats of Alabama in a year
at least 300 years into the future. They might not be in USA however.
I failed to define that. It is moot that details are appearing so
slowly even though Rob Richie appears to have concluded explaining
how it was that IRV was exonerated.

Mr Richie just asked "why ... [?]" and addressed the message to me,
therefore there was not a question to me. Check the document. Anyway,
it is plain to me, from the wording, that response to the question
I got, is not used. It looks like Rob was hoping to have the
argument fail to exonerate IRV if he and I agreed over some issue.
That is obviously a use of wrong reasoning to come to a maybe
correct conclusion. What exactly it was that was causing IRV to be
found worse than ever in 3 candidate elections was not clearly given.

Partly it was that I seemed to have found he worst case example and
so that could mean the next step is 4 candidates. It could take a
PC maybe as much as 4 minutes to find the next worst case example
(if not simplifying and the form is rightly guessed at).


Here is a use of a question about IRV where a question is asked and
then the answer is used: some referenda where the [renamed]
Alternative Vote is chosen:


This is how the example was first found. It is just 6 lines of code that
did the easy quick calculations:

a_w1 = r:=(b<a)and(b<c)and(c+bc<a+ba) or (c<a)and(c<b)and(b+cb<a+ca)

 >a_w1 := meth(stv, 'a, a0,ab,ac, b0,bc,ba, c0,ca,cb);
 >e1 := (0<x) and a_w1 and nott
 >· sub({a0 = a0 + (1-f-g)*x, b0 = b0 + f*x, c0 = c0 + g*x}, a_w1);
 >e2 := nott rlqe ex ({x, a0,ab,ac, b0,bc,ba, c0,ca,cb}, strict e1);
 >e3 := rlqe all ({f,g}, ((0<=f) and (0<=g) and (f+g<=1) and (z < 1-f-g))
 >impl e2);
 >% REDLOG returned these values:
 >% e2 := 2*f + g - 1 <= 0 and f + 2*g - 1 <= 0 and f - g = 0
 >% e3 := z - 1 >= 0
 >% An example is found (using "rlqea rlex sub"), especially since the f-g=0
 >% is the sort of thing that could appear due to a mishandling of surfaces.
 >· · · · ·-----------------------
 >· · · · ·A· · 49,000· · 50,940
 >· · · · ·B· · · · ·0· · · · 20
 >· · · · ·BC · 25,505· · 25,505
 >· · · · ·C· · 25,495· · 25,535
 >· · · · ·-----------------------
 >· ·Winner:· · · · A· · · · ·C
 >· ·Total: · ·100,000 · 102,000
 >Thus the Alternative Vote allows the addition of the papers
 >· 2,000 * (97%:(A) + 1%:(B) + 2%:(C))

to harm candidate A.

  From: http://groups.yahoo.com/group/politicians-and-polytopes/message/164

Had Mr Richie actually won the argument where we got to learn so much
about the actual principles he would sometimes use, then that 2nd
example was a back. I did not see any perfectly exactly idea in Mr Richie's
argument that would allow two of these 2-point examples to be devalued
at the same time.

Lets actually do that: I argued that the first example pair showed a very
bad problem with IRV and maybe there was a suspicion that IRV had being
badly faulted. Just for the moment, suppose that Mr Richie was victorious
in the previous argument. Beyond all doubt, Mr Rob Richie would be no
less concerned with the example above, which never was countered with a
"look elsewhere" argument. The CVD would not be defenceless to point
attacks if it had of given the principle behind that "look elsewhere"
style argument. At my PaP, the readers get to read quite a bit about
principle. If the IRV Executive Director had the ethics of a computer
scientist building equipment to affect lives, then each attack could or
should prompt a search. Anyway, it can only tend to lead to harm to
politicians and those people probably isolated and stuck for reasoning
if injust systems for them are being planned.

Now that the policemen of unfairness, the CVD, have divulged that they
have no respect whatsoever for equal suffrage and maybe P2, then it
will also be the case that the CVD should not be expected to do research
into methods that are not fair at all. Thus Rob Lanphier's group here
produce better methods, e.g. approximate methods. That is raising the
bar too high. Maybe just wrong rejectible methods. Why is Schulze's
method rejected?: is it truncation resistant?. The question would be
asked again if not answered. Demorep did not answer that question.
That is where truncating after a given candidate's preferences does
not change the win/lose status of that candidate.


For those that missed it. there are billions of better one winner
methods than IRV.

I can roughly plot methods between FPTP and the so called IRV method:


    |                                          |                     |
   FPTP                                       IFPP                 AV/IRV
    |       Good monotonic methods            |about here|           |

It may take many methods before Americans get to see a good method being

A "good monotonic method" includeds the [so called?] Cumulative Vote method.
No one answered my question so I ask again: it this the Block Vote and/or the
Cumulative Vote:

Paper (C): Add: a=+0, b=+0, c=+1
Paper (CA): Add: a=+1/2, b=+0, c=+1/2
Paper (CAB): Add: a=+1/3, b=+1/3, c=+1/3


Maybe Mr Forest Simmons may reply to this.

Where can IRV and/or the CVD, have their point plotted in here:

 >         The assertiveness
 >        of the claim, truth.
 >                /\
 >               /  \
 >              /    \
 >             /      \
 >            /        \
 >           /          \
 >          /            \
 >         /              \
 >        /________________\
 >     IRV is              IRV is good enough to be
 >   good enough            desired and its good
 >   and passes              are found by considering
 >  just as cleanly as        normal vectors alone.
 >  a metal engineer's nut     How the IRV method is
 >  would pass through a        the material expression
 >  larger precisely            of the desire of the CVD
 >  measured hole               to advance their knowledge

I'd delete the lower right point first.

So I guess the CVD's advocacy of IRV ought be positioned right at the
very top where truthfulness is measured, after noting that the CVD
somehow ends up saying little of mathematical precision so it is not
near the lower left. The lower right corner is adverse to sleazy
corruptnes and the gross vote negating defects IRV exhibits.

It has to be truth that is the hallmark of IRVism since that test
is nondemanding, as worded. I got the bottom two wordings to be
strict and it is sort of crapped since the CVD's capacity to
understand its own ideals if vague. The top point is mislabelled.

Anyway the comments of Mike can lie exactly on the top point because
his messages are much more undefined than provably untrue.

There it is: Mike is just like the CVD.

Help me with the Section 502(c) text. Is that an IRS number or a USA
Titles and Codes number?. The CVD about page has no reference. They
simply beg for the money and didn't want to waste space on saying
that they were correctly registered and under which statute.

The Cumulative Vote (receiving AV/STV/IRV papers) passed perfectly all
the equal suffrage (P4) testing that I have so far done of it. It can't
be found corrupt. I just mention that. No power excesses of size 1666
if the view is limited to only 2 FPTP votes.

My last message has lots of errors in it.


                Better Methods From Ossipoff.

Mike can't write questions. I can fix them minutely and it is doubtful
that that is going to be acknowledged. Mike is behind that he might as
well learn from me.

PS. Mike is like Mr Richie in that they both use that strange phrase
for the good English term, the Alternative Vote. It is simple to fix
there Mike, just change your terminology. It is sharply alarming to
read that you were a member of the CVD. That means that every
statement made may have special types of flaws that havve yet to be
noted or discovered. E.g intentional bias when the facts do not 'call
for that'.

A tiny problem: the ideal.  Here is a question:

At 2002\03\25 03:13 +0000 Monday, MIKE OSSIPOFF wrote:
 >Maybe much improvement could be gotten by methods that use more
 >than one balloting, especially more than 2 ballotings. Maybe
 >idealness can be approached more if methods additionally use
 >more than 1 kind of input.
 >How good could such methods get, in regards to the goal of
 >reducing strategy incentive or strategy dilemma? Which method of that
 >type seems the best in that regard?
 >For 1-balloting, nonprobabilistic methods, is there any method
 >that can do better than Approval, CR, & Condorcet(wv) when judged
 >by that standard?

Can Mike count the number of variables and parameters to any of the
rules of Mike. Maybe no one has seen a rule of Mike, i.e. in an
existential logic quantifier format.

As members may recall, Mike's rules were only ever written in
English and they had top secret input channels in addition to the
sets returned by the 'winners' function(s) and all the variables
that the instances of the function used.

Mike would not count the number of variables and disclose the
count of the number of parameters and so it seemed that they were
secret from Mike too. Also none of that led to any progress as I
recall. It looks like it may happen again since words like
"incentive" are hard to interpret in the absence of voters.

When the aim takes account of voters then there is either nothing
(e.g. nonsense) or a real constraint that is going to reduce
proportionality or else something else.

Quite simply there is not a need for a search for what makes a
method sincere. It is just proportionality. The whole statement
of an intent to advance knowledge is drafted badly enough that
it might as well be made something that is forgotten. E.g. the
main difficulty for finding the rule falls on the word "goal" and
strict rules are not goals.

Until Mike All the principles that Mike might want are online
at my http://www.ijs.co.nz/ifpp.htm webpage. That is as good as
Mike's electionmethods.org or whatever it is, is bad.

Mike is trying to distress the EM List of Rob Lanphier who may
be determined to not learn anything much at all about a better
preferential voting method based on axioms. Actually, Mike did
not say that the method would be based on an axiom. If there
is a redraft then that could be fixed.


I am unsubscribing from the list. I can't hold a presence down
here for too long. Nor can members quit and come to my PaP list
since that is on 4 candidate problems and this seems to be on
2 candidate preferential voting methods. It seems to not
produce 3 candidate algebraic (polytope) equations. It is like
there should be a 3rd list. It could a decade discovering what
I posted into PaP 1 to 2 years ago, but rejecting Condorcet's
views might not happen. People here believe in pairwise
comparing and it is obviously wrong and then nothing sort of
comes from that. I ask Ossipoff: why did you say you wanted
a better method than Approval and Condorcet when not constraining
the number of kinds of papers to be under one thousand million ?.

PaP:  http://groups.yahoo.com/group/politicians-and-polytopes
STV:  http://groups.yahoo.com/group/single-transferable-vote
Preferential voting axioms: http://www.ijs.co.nz/ifpp.htm

To Forest or someone: please comment on IRV's corruptness,
when there are 3 or 4 papers only.

Just testing being a reformer here:  "You will agree with me obviously".

Craig Carey

More information about the Election-Methods mailing list