[EM] Voting Matters #19

[Craig Carey]

The paper starts out by withholding the definjition of the numbers
that are used in the (believed to exist) pairwise comparing graphs.

There is something too stupid for the author to put into his paper.
I ask in this message and in the message at the STV maiing list,
for Mr Green-Armytage to tell me why he found it essential to
guarantee that whatever theory he has got is unfair.



At 2004-11-28 23:20  Sunday, James Green-Armytage wrote:
>
>Hi folks,
>	As some of you might already have gleamed from Craig Carey's ravings, the
>new issue of Voting Matters came out last week. I wanted to do a short
>post calling attention to my article in the new issue, and inviting
>everyone to read it. The article is entitled "Cardinal-weighted pairwise
>comparison." It describes the cardinal pairwise method, argues that
>majority rule methods should be Condorcet-efficient, and argues that the
>cardinal pairwise method will be more strategy-resistant than most other
>Condorcet-efficient methods. I worked hard on the paper, and I'm fairly
>proud of it. Here are the links:

The paper claims that "cycles" can exist. However because of how he failed
to define the numbers of the arcs, when Mr Eppley or Heitzig or some other
enemy of fairness, says that the cycle is clockwise, it might in reality
actually be anticlockwise once that all the secrecy on how to compute the
numbers is undone and Shulce "d" matrix values that chance to be much
less unfair, are used instead.



>Voting Matters main index page
>http://www.mcdougall.org.uk/VM/MAIN.HTM
>
>Issue 19 main page
>http://www.mcdougall.org.uk/VM/ISSUE19/INDEX.HTM
>
>Issue 19 pdf
>http://www.mcdougall.org.uk/VM/ISSUE19/ISSUE19.PDF
>
>Cardinal pairwise paper pdf
>http://www.mcdougall.org.uk/VM/ISSUE19/I19P2.PDF
>
>

I wrote on the message of Mr Green-Armytage:

http://groups.yahoo.com/group/single-transferable-vote/message/332
| Message 332
| 
| From:  Craig Carey <research at ijs.co.nz> 
| Date:  Sun Nov 28, 2004  8:39 am 
| Subject:  Voting Matters 19: Five secret weighting numbers again!: fairness
|                                                         vs pairwise comparing


It looks like everyone can pop down to Voting Matters shed a skin of
lies. Last we got heavy clues at the start of the VM 17 article that
Mr Schulze had perhaps starte. I am too intelligent to learn anything
at all from Mr Eppley. The man needs a lot of tuition since a while back
he exposed to us the disgraceful blubnder of presuming that there is
something to be respected in the Condorcet winner.

Sure: in a Cambridge election with 200 candidates, Mr Eppley and
Mr Grere-Arymtage expect that ever (AB) paper that is added has got
a good chance of killing off candidate 'B' by reducing the "A over B"
sum, even if A is a complete loser. I can't write "or a complete winners"
since unlike with fairness, the entire Condorcet belief is apparently
totally useless if there is 2 or 0 winners. An extraordinarily long
apology is expected, and Mr Green-Armytage can round up a list falsifications
and explanations and post them to the fair-minded single-transferable-vote
mailing, if the secrecy on who thought what is not excessive.

I suppose MIKE OSSIPOFF could lurch with a forgetfulnerss of why it is
so reliably the case that those numbers are not defined. Mr Schulze
did the same and added to his intent to not actually define any
preferential voting method, some praise of it.


I leave to readers to pick between these two:

(A) Almost all randomly constructed methods are failed by fairness rules,
  particularly when monotonicity is checking while there is over 30
  candidates. Perhaps Mr 

 For example, having the candidate-A-wins Boolean expression contain
  the (b<c) flat at a surface, can be known instantly by even a dimwit
 (and they like pairwise comparing here), as being something that will
 cause the method to be failed by a monotonicity check.

 Perhaps Mr Green-Armytage didn't know the voting polytopes can be
 simple written down as an expressions.

 The term would not offend against the monotonicity rule if it was
  (2*b<a+c). Any admission of Mr G-A saying he never knew that would
  indicate that all of his articles at Voting Matters might be ones
  that the public can avoid reading while missing nothing.

 To have a (b<c) term visibly inside of the A-wins equations (and
 exposed in a face) will allow candidate A to start winning when
 (A) papers are changed into "(C)" papers.

How can Mr Green-Armytage get his pairwise comparing method succesfully
through a simple fairness tests ensuring that the piower of all papers
is between 0 and 1, whilst never having made any attempt at all, to get
the whole theory, or whichever method, past the test.

That test that he fully ignores -- embodying some essential human right
to have a fair deal, etc. -- is imagined by Mr Heitzig to not exist.

It is really simple: if the other requiers that A-wins contains
(b<c) then the entire theory of Green-Armytage is possibly garbage and
normally every single reader also misses out an excuse hinting that
a theorist entered into a struggle on behald of justice and somehow
lost.

Alternatively if "b<c" is written as "(2*b<a+c)" the Mr Green-Armytage's
entire belief system survives a little longer. So far it did not
survive until the time when he explained that he did a lot of work.

I don't know what the purpose is, for sayiong that it is hard work
to be unfair to whomever and word up the paper to that the evidence
of that is missing. Naturally that would trash the credibility of the
entire article, but I assume Jobst Heitzig has been imagining that
he can outdo Eppley, Schulze and Mr G-A in keeping secret why he
got the weighting constants wrong and the whole design of the
weighting method wrong.

Will Mr Heitzig tell us if he would ever do what Shulze and Mr G-A
did, and keep secret the first step that processes the ballot paper
counts ?. I assume that Mr Heitzig can't actually privately or
publicly with any person is prepared to say that even German voters
have a right to a fair voting method. Probably a man for each city a
and the "odious apparatus" of the tyrannical forces that place
desire before reason. But instead we get treated to show of secrecy
and no definition appears in a English PDF that discloses a definition
of what happens to votes (i.e. counts) in the first step.

Give us your name, and if you are not a family member or if you are
indeed an American you can all line up and let me know what your
best guess is, for how the "A over B" numbers are computed.

Voting Matters might have an English audience so it can spot total
lies about where to put your 5 categories:

(..A..B..)
(..B..A..)
(..A..)
(..B..)
(...)

The opponents of Mr Eppley's and Heitzig's false belief saying that 
airwise comparing is of value, would have the correct and
perfectly exact alternative idea of:

   making no statement at all.

Can anybody name a person in England who make the same error that
Mr G-A made, and guarantee that method is unfair since using the
(even today, still undefined for the planet's public) "d" matrix of
Mr Marcus Schulze. We can see what the big problem is for Mr G-A,
he expected to let go of the lies he soaked up from this mailing
list and he explains that he worked whereas any competent method
designer would replace the worthless idea with nothing. There is
no hard work in making zero inferences when starting off with the
fairness-compatible axiomatic stance of saying nothing.

I suppose it is true and Mr G-A did work hard, However the topic
is why he is the second person to keep secret the first
data processing that almost guarantees that the method is unfair
and no government will use the method. Of course G-A's ideas would
end up being complete failures in black-box fairness testing. What
else would anyone expect. It is a bit excessive to conceal the
first steps in the copyable human activity of getting an unfair
result that is just some exploding mini-universe of complexity;
i.e. his 1-winner polytopes has perhaps thousands of times more
faces than fairness would lead to. If fairness is only able to
explain 0.1% of all the faces, then why exactly is the purpose behind
the other 99+% all kept partly secret in the write-up.

If we have not already passed it, then the future may hold some
central communist committee meeting of pairwise comparing devotees,
who will formulate a policy on what else to keep secret.

For example maybe Voting Matters will start to reject pro-Condorcet
aritcles of EML morons if those elite few they don't start
concealing the fact they were scavenging around for stupid rules
that somehow pass their method and a rubbish tin of worthless
alternatives and some randomly created method.

To beat Mr Gilmour to the line, persons in England will later be
expecting that rules be subject all sorts of checks including those
that are desired to be done.

Things are so dumb here are EML that seems possible that every PDF
document creator is victim of a bad habit of leaving the X over Y
numbers undefined. Mr OSSIPOFF can write anything here: for or
against secrecy. For 4 or 6 years OSSIPOFF has kept secret from us
the algbraic formulations of his beliefs: it seems excessive to
hope that MIKE OSSIPOFF would write "b<c" before January.

Those central committee communist meetings of what to keep secret
from the public, might also be perceived to be a hard slog for
Mr Schulze and the other believers in Condorcet.

It is all like Mr Schulze forgetting to mention that planet might
disappear in a few years due to some galactic core that materialized
and that is now ripping through the solar systems.

I.e. the incredible complexity of the downstream implications of
Jobst Heitzig's beliefs are being kept secret too (along with the
first step that receives ballot paper counts).

How does the German and Mr G-A. do it, hide the horrendous complexity
that is expected to be absent in pairwise comparing theroies and
present in theory source from Condorcet devotees ?.

Of course logic equations defining who actually wins are always
never supplied by a Condorcet believer.

If I ask Mr Heitzig, the non-mathematician, this OSSIPOFFianly
simple question:

    WHEN DOES CANDIDATE A WIN ?

then Mr G-A or Heitzig would doubtless consider providing a bit
of an algorithm. I NEVER asked for any algorithm or just the
fragment of the algoring that might get released. 

Even if say that IRV has a 2 line long equations, Jobst Heitzig is
probably quite likely to conceal from us which of these 3 is the
truer:
  * the algorithm is 20 lines long;
  * the algorithm is 50 ines long;
  * the algorithm is 100 lines long,

For 2 line algorithms an (almost certain) failure under monotonicity can
be detected after mere seconds of looking at the algorithm.

For the pairwise comparing theorists the most likely situations is
the pairwise comparing students keep complaining that the single
expression for any of them to actually produce. 

I identified the problem a quick fix not taking more than 4 years was
to create a new symbolic algebra software that can have a few
Clifford/Grassmann algebra non-commutative features. However at the
Election Methods List every student of Mr Eppley or whoever has to
bluff their way past the question asking when candidate A wins.

Mr Shulze was writing on the Floyd algorithm and he did not mention
that the data going into the algorithm was unfixably wrong.

I assume that the American students just soak up false beliefs.
Things could be seriously wrong in the mind of the questioner if asking
for the strict logical reasoning for that wrong belief instead of why
the students desired that.

Perhaps the most surprising thing that could happen here is for the
creature of HOTMAIL,COM to do what I expected of one Jobst Heitzig, and
post up a full symbolic algebra equation answering the question of when
candidate A wins a 2 winner election, e.g. the 2 candidate 2 winner
election. 

The discussion could be transferred to the politicians-and-polytopes
mailing list. However the British would simply pick out, and throw
away, an the wrong belief that if (AB) + (AC) papers are added then B has
a bigger chance of losing. Certainly everyone that believes that seems
to be quick reconfigure their e-mail filters and have no intelligent
response. This is classic symptom of a mailing list run by Rob Lanphier:
freefalll in standard and less than one chicken;s grain of research into
fairness s every 20 years or whatever.

I have a mailing list where questions are actually answered. I guess
that there is a bit of a chance that that was why Mr Green-Armytage
wanted to get the comments transferred to this mailing list for
Americans who lack the security passes and who are not entitled to
know. It is more interesting now that a 2nd person did exactly the
same and concealed the whole idea of what a preference number is.


We could do with a whole new list of references to pairwise methods
that are actually defined.

Mr OSSIPOFF once wanted "understanding" from Americans. E,g. the voters
are who "YOU" want to be.

Now we the collective understanding of devotees of pairwise comparing
who tumble over each other (words describing sheep) when trying to
keep secret that they can't get the numbers labelling the graph arcs,
right. Mr Heitzig is as careful to keep that secret, isn't he ?. I have
not been reading so much dull material that leads only to failure.

Of the many fairness rules, monotonicity is interesting in that
an extremely rapid opinion on whether a method is failed by monotoncity
can be formed just by looking at the X-wins logic expression.

The equations of when A-wins are missing, I imagine that OSSIPOFF
desires to criticize people like Marcus Schulze of Berlin, for actually
failing to post in one of those logic expressions that get exactly the
same result as an equivalent algorithm. With the Germans charging
many for their BMWs and etc, we really ask these 2 Germans what price
would have to be paid to the solidly-coalesced-into-unfair-beliefs
duo. Perhaps USD$50 might be enough to get Mr Heitzig to produce the
first 70 lines of his favourite A-wins eqations. The competing
Alternative Vote could be only 2 lines line and probably less of a
failure under the rights testign rules that themselves were tested.


Mr Green-Armytage had a different vision: rename a rule or everlastingly
good axiom, a "Criteria". That was a dumb trick of the www.condorcet.org
man who quite explained privately that he was not responsible for the
(apparently unmissable low intelligence of this mailing list. Anyway,
things probably get worse from here. I suspect that the owner also does
not do research so it is up to the American university Condorcet believers
to redouble their efforts to appear to be too confused or clueless to
achieve something. Mr Ossipoff tries to help the CVD succeed, by keeping
secret the Twin Towers Cascading Nobody example demo. Instead MIKE
seems to make his best arguments be wrong arguments that fail optimal
correct methods in addition to the IRV he does not like. Any year he
corrects that wrong belief can't come too soon.