[EM] EML's failure to understand 2 candidate maths
Craig Carey
research at ijs.co.nz
Tue Jan 11 04:56:03 PST 2005
Mr Marcus Schulze has failed the challenge that I set.
The method that is to be derived is simply disclosed before the
person taking up the challenge, starts.
Mr Schulze's failure seems to be what I would expect if he had a
purpose of concealing 5 weighting numbers or something. I assume
subscribers have are aquainted with how Mr Schulze is entirely
confident when not disclosing the secret sum denoted by hist
words "[the number that] strictly prefer [X over Y. etc]".
---
I can't tell if Mr Schulze's "strictly prefer" defines the first lot of
weights or the second. Here are the two.
[Group 1]:
Total for A: 1,1,0,0,0
Total for B: 0,0,1,1,0
or [Group 2]:
Total for A: 1,1, -1,-1, 0
Total for B: -1,-1, 1,1, 0
All of the ballot papers are these:
(A) a0
(AB) ab
(B) b0
(BA) ba
() z0
Unfortunately Mr Schulze seems to avoid voting research lists and
wont write privately. One of the few private messages I got said
he would answer questions. So I simply asked for the 5 weighting
numbers and the sum, and Mr Schulze welched on the promise and I
never got to see a definition of Mr Schulze's "strictly prefer"
sum.
I didn't expect Mr Schuzle to fail so quickly in this challenge.
I was unsubscribed but the writing marred with the mistake of
Mr Schulze presenting himself as a man who was succeeding in my
challenge to prove that there is some competence in 2 candidate
mathematics.
Here is the obvious 2 candidate 1 winner correct preferential voting
method:
(A wins) = (b0 + ba < a0 + ab)
(B wins) = (a0 + ab < b0 + ba)
The papers are the same five as above, i.e. this:
a0*(A) + b0*(B) +
ab*(AB) + ba*(BA) +
z0*()
That method is definitely not First Past The Post.
----
In about 2001 ?, Mr Shulze once said that he liked IRV.
IRV also provides the correct 2 candidate answer.
Followers of IRV might like the Truncation Resistance axiom. It is
just a hint on how to solve the 'challenge'.
Having no comment on Truncation Resistance is a trait of this
mailing list. Worse is described below: junk definitions of Mr Eppley
and others that serious flawed: e,g the definition of monotonicity.
---
I make a quick catalogue some of the recent detected mistakes of
Mr Shulze, who I judge to have failed the 2 candidate challenge:
(1) He talked about FPTP/FPP. It is was one method of an uncountable
number. He picks at random, or guessed that economists like FPTP/FPP.
He can't solve the problem by selecting randomly. Selecting a
favourite of unnamed economists is another way to fail to produce
a quick simple winning response to my 2 candidate challenge.
(2) the question was asked just *after* I had answered it. What
happened was that he asked for the information after it was given.
To have run out of good arguments so fast, and to have used
repetition, leads me to close the challenge and find that Mr
Shulze had failed. It could be debatable.
(3) He saw the 5 paper solution because I wrote it out using
the symbolism of algebra. I am not quoting his text I can't be sure
what he was trying to say. I suspected he was being ambiguous
on whether 2 candidate (1 winner) First Past The Post handles
only 3 papers, or only 5. A correct answer is "only 3 papers".
(4) He tried to say that I rejected FPP/FPTP. I do since that
has no connection with his argument.
The request was axioms and reasoning based on the axiom that
allowed the method (that is Not FPTP/FPP) to be decduced.
(Unless somehow an inductive argument can be deduced. It may be
more safe to say that the given solution is "derived" or "inferred").
(5) Mr S copied undefined wrong ideas of economists into the list.
Most of them are not worth memorizing. He didn't identify any
particular economist's idea as an axiom that he wanted for the
logical deduction that was required for the challenge.
(6) Mr Schulze seemed to be close to saying that the 2 candidate
First Past the Post method handles 5 ballot papers. I could not
decide on what he meant. But FPTP/FPP doesn't.
Maybe a vote would back me up.
---
A while back Mr Schulze made a wrong claim that his Schulze
method was monotonic. That appeared at Voting Matters.
I might send in a letter to the editor correcting the error.
I expect that Mr Schulze won't admit to making a wrogn claim
about an undefined method. I don't do that myself. Mr Schulze
should back out of the error. Good theorists do not make errors in
the topic of mathematics.
I might wait a few years.
----
Mr Heitzig would like to have the Debian project get the
winners wrong. Anyway that is not my topic here.
Mr Heitzig's River method would be unfair but I suppose he
leaves it up to me research into the thing. The other man
withheld the definition of his algorithm.
There is some music by Schubert with grat new cadenzas,
at the River BBS (and in CMF format instead of Midi format):
The Wanderer: http://riverbbs.net/files/output/40701.Html
CMF to Midi: http://www.hitsquad.com/smm/programs/cmf2mid/
---------
I saw that Mr Suter didn't say that the CVD has a first aim
of under-promoting the 1/3 quota when 3 or more candidates.
There is something fishy about how men in USA don't mention
the most important thing first. What does the CVD actually
offer and what is the best that it could offer.
By rejecting fairness, the answer can be wrong.
The FORTRAN language is being improved: FORTRAN 2003:
http://www.kcl.ac.uk/kis/support/cit//fortran/john_reid_new_2003.pdf
----
The non-existent 5 paper FPTP/FPP of M<r Schulze can be obtained
by defining ** proportionality **.
Suppose the papers are these:
(A) a0
(AB) ab
(B) b0
(BA) ba
() z0
Then the subtotals are these:
Subtotal for A = a0+ab+ba
Subtotal for B = b0+ba+ab
There is 1 winner. The candidate with the largest subtotal is found.
And so:
(A wins) = (b0+ba+ab < a0+ab+ba)
Simplifying I get: (A wins) = (b0 < a0)
That is very similar to First Past the Post.
Unlike Schulze's First Past The Post, it accepts 5 ballot papers
instead of 3.
Mr Schulze can missing simple interesting little details like that
via the improper trick of telling us about old worthless ideas of
his economists.
----
I need to produce some finishing observation.
I might try this:
We have first contact with the Ctenophore theorists.
| A ctenophore is marine invertebrate (lacking a backbone).
|
| The sea gooseberry and "comb jelly" is a ctenophore. The word
| comes from the Greek word: kteno-, meaning "comb", and the
| the suffix, -phore, is added to that]
Mr Schulze failed to identify an axiom.
Actually, I guess I would most like to hand out my Ctenophore
award to Mr Russ Piaelli, areospace expert, who programmed up a
Shulze method algorithm that incredibly accepts a seemingly really
undefined pairwise comparing matrix, instead of counts of STV
style ballot papers.
Some of the algorithm is below.
---
I have never seen Mr Schulze talk about fairness.
There is no evidence that the 2 Germans harmoniously accept
each others (wrong) beliefs.
Or the Tannhauser overture to remind of ascent and descent.
But I get to see little mistakes. Normally such a large quantity
of mistakes that I am offline and not even subscribed to the EML
list.
I knew theat Mr Schulze does not use reasoning. However the challenge
was winnable with perhaps 3 lines of argument. Mr Schulze might
have lost interest after failing.to deduce a 2 candidate method.
However there are other people who can be checked (examined).
---
** Mediaeval monastery libraries : how to not check **
Here is the webpage of Mr Steve Eppley who occasionally writes:
http://alumnus.caltech.edu/~seppley/
This definition of Truncation Resistance is too different from the
agreeable definition.
| TRUNCATION RESISTANCE: Define the "sincere top set" as the smallest
| subset of alternatives such that, for each alternative in the subset,
| say x, and each alternative outside the subset, say y, the number of
| voters who sincerely prefer x over y exceeds the number who sincerely
| prefer y over x. If no voter votes the reverse of any sincere
| preference regarding any pair of alternatives, and more than half of the
| voters rank some x in the sincere top set over some y outside the
| sincere top set, then y must not be elected. (This is a strengthening of
| a criterion having the same name promoted by Mike Ossipoff, whose weaker
| version applies only when the sincere top set contains only one
| alternative, a Condorcet winner.)
The term "sincerely prefer" is not actually defined on the webpage.
It reminds of Mr Schulze never does not define the words "Strictly prefer".
One plain error is in the words:
"y must not be elected".
Beginners can avoid that sort of mistake: suppose there are 2 winners
and 2 candidates ?, or 5 winners and 5 candidates ?.
Obviously Mr Eppley was not running essential and trivial checks over
his ideas.
Since "sincere" was undefined, I searched for the word "probability".
Both can occur when there is fantasizing that voters are around the
corner in large crowds as far as the eye can see, and seconds away
from cheer uncontrollably about the genius of the EML theorist.
This text on the webpage contains the word "probability".
http://alumnus.caltech.edu/~seppley/
-----------------------------------------------------------------------
| MONOTONICITY (non-negative responsiveness): If some voters raise an
| alternative in their rankings, then its probability of being elected
| must not decrease.
-----------------------------------------------------------------------
Note that the motive for using the word "alternative" instead of "candidate"
seems suspect. The word "alternative" woudl naturally mean "set of winners"
However it is bent to take the unnatural meaning of "candidate".
The same fishy wording is to be found in Mr Cretney's definition of
monotonicity:
http://www.condorcet.org/emr/defn.shtml
-------------------------------------------------------------------------
} MONOTONICITY
} The property of a method where an alternative can never be made to
} succeed by being ranked lower on some ballots. Doing this is using the
} "push-over" strategy.
-------------------------------------------------------------------------
It could have been "social choice function" which is worse.
Anyway a severe & major error in Mr Eppley's definition, is to requires that
support rises help, even when the papers have constant negative counts.
The monotonicity definition was bungled. Thus seems to show that followers
of the pairwise comparing belief can bungled their most important
definitions and never fix the mistake even if 5 years pass.
But it seems that the word "some voters" is doing unnecessary discretizng
of the counts.
If I wrote this:
"(Exists x, x is an Integer) F[x]"
then there is some quantity of "x" and that quantity can be negative.
Just to clarify things, let me say that there absolutely are not voters
and the task for us is to come up with a meaning for Mr Eppley's
words "If some voters [etc.]". Clearly he sought have the rule say
nothing when there are no voters and there is a method to be tested,
It is another major error on the rule that could not possibly be put
in there by accident. It is seems likely that one of the two men,
copied from the other.
---
Mr Cretney's uses the word "lower" when meaning 'higher'.
(The 1st preference has a lower rank than the 2nd preference.)
However Mr Cretney has seen these arguments in the past.
Mr Cretney's idea of "[can't] succeed" will be badly malfunction when the
number of winners equals the number of candidates.
It could that yet another believer in Condorcet is incompetent in the
mathematics of 2 candidate elections.
I have not seen reasoning for why very many mistakes exist,
The 2 men could have copied from each other.
Mr Schulze seemed to be wildly copying ideas of economists just
prior to failing my easy challenge. The members of the
politicians and polytopes mailing list would not wildly copy.
----
Mr OSSIPOFF seems have an online copy of the Schulze method.
Like Mr Schulze, MIKR OSSIPOFF left the start of the algorith,
very undefined.
The webpage is here:
http://www.electionmethods.org/Condorcet-alg.htm
You can catch the whole Schulze method in the Python code
http://www.electionmethods.org/CondorcetSSD.py
There is no reference to Mr Schulze.
Here is the name of the method:
__________________________________________
The "Condorcet Cloneproof Schwartz
Sequential Dropping (CSSD) algorithm".
__________________________________________
It is a reality of modern Condorcet variant designers:
they just don't seem to feel in the mood for defining the
first input stage of their algorithms,
It could be a small topic.
They reject fairness in all of its many forms. That is not explicitly
admitted. A core part of fairness is monotonicity, and they seem
to exchange unusably bad definitions.
The leading nose cone of this Schulze algorithm by Mr Paielli is missing.
No person would build a fast plane like that: have no nose cone so that
shock waves pull thje metal off as if jelly was being cut with a metal
comb.
|
| #!/usr/bin/env python
|
| # Copyright (C) 2002 by Mike Ossipoff and Russ Paielli
|
| # version 1.0 - released 2002-02-14
| # version 1.01 - released 2002-08-24 -- comments revised slightly
| # version 1.02 - released 2004-02-14 -- comments revised slightly
|
| # See http://ElectionMethods.org/CondorcetSSD.py for updates.
| # See http://ElectionMethods.org for related informatioin.
|
| # This python script implements the Cloneproof Condorcet SSD (Schwartz
| # Sequential Dropping) voting algorithm and provides associated
| # input/output utilities. The Condorcet SSD algorithm is known to be
| # equivalent to the Condorcet Beatpath Winner algorithm, and we consider
| # it the state of the art single-winner voting algorithm. Mike Ossipoff
| # provided the algorithm, and Russ Paielli programmed it. To use it,
There is no mention of Mr Schulze in the text.
I am pleased that Mr Schulze is not mentioned since he withheld from me
his algorithm (starting in 2003).
However Mr Paielli and Ossipoff's write-up has a similar problem if
only the 2 named URLs are checked,
| # type
|
| # CondorcetSSD.py <input> <output>
|
| # where <input> is the input file, and <output> is the output file. The
| # command-line arguments are optional. If <output> is omitted, the
| # output will go to the standard output. If <input> is omitted, the
| # input will come from the standard input. A valid input file specifies
| # the pairwise voting matrix in the following format:
|
| # .SUM Smith: 0 44 83
| # .SUM Jones: 49 0 29
| # .SUM Brown: 93 75 0
|
| # Note that the ".SUM" keyword must start in the first column, otherwise
| # spacing and column alignment is arbitrary. The column ordering
| # corresponds to the row ordering, of course. For example, Smith got 44
| # votes over Jones, and Brown got 93 votes over Smith in the example
| # above. This format is the standard output format of GVI: The Graphical
...
I would prefer an antisymmetric "X over Y" input matrix.
But the important thing is that the definition is missing.
I could hunt around the OSSIPOFF website and see if he created
the Matrix that is input for the Sequential Lamb Droppings algorithm.
|
| from sys import exit, argv, stdin, stdout
|
| def CondorcetSSD ( Pmat, debug=0, output=stdout ): # Condorcet voting
| # algorithm with Schwartz Sequential Dropping (SSD)
|
| # Pmat: pairwise voting matrix (input)
| # Pmat[i,j]: number of votes ranking candidate i over candidate j
|
| nc = Pmat['nc'] # number of candidates
|
| # Initialize the defeats matrix: Def[i,j] gives the magnitude of i's
| # defeat of j. If i doesn't defeat j, then Def[i,j] == 0.
|
| Def = Pmat.copy()
|
| for i in range(nc):
If MIKE OSSIPOFF knew the definition of the Schulze "strictly prefer"
matrix then he would have put it into the Python algorithm.
-----------
Back to the challenge - Any more ready to show deduction?
Mr Venkze and Mr Suter and Mr Eppley and others can provide their
own solutions to the new challenge.
If any can solve a 2 candidate method then now is a good time to
make that known to the public.
It seems that believers in Condorcet ideals make too many mistakes.
Perhaps former CVD members can do better (e.g. MIKE).
I keep reading about how the CVD was formed in 1992.
Mr Eppley seems to avoid the "if wins here then wins there" style that
fairness rules have. Why would Mr Heitzig not talk on fairness:
surely the results of fairness are much more attractive to politicians
than whatever his River algorithm is. Mr Heitzig's is keeping a
secret just as Mr Schulze was. The 4 candidate A_wins logic expression
could fill up much more than 1 A4 page if written out
A quick fix (built in Germany) is to avoid using the word polytope.
It should be possible to mislead masses of students in USA by
never actually correctly naming the object being studied.
A 4 candidate "IRV" variant might have a complexity similar to that
of a 3 candidate Condorcet method variant. There has a been a
concealment of that. So better methods from Mr Heitzig might not
be better (since he does not call polytopes polytopes and rejects
fairness and accepts wrong rules I guess. Too many unpromising
aspects here).
Mr Lanphier would say that the list is running brilliantly. He has
an interest in keeping standards low, or else USA students could
simply quit faster.
I run a mailing list for 4 candidate methods.
I guess Mr Schulze gave up on the 2 candidate method. I see that
Mr Schulz is not a member of the politicians and polytopes mailing
list. He presumably finds it unsatisfactory in some way.
With the incomprehensible first ranking Condorcet method
designers creating rules that malfunction badly in 1 candidate
elections, it is unsatisfactory to let that be ignored.
Perhaps the USA subscribers could split into teams and try to solve
the problem of deriving the 2 candidate 5 paper 1 winner algorithm.
[PS. This message is at the start of another new thread, since my Eudora 5.2
program does not allow me to add the needed headers. And Eudora does not
thread messages too]
Craig Carey <research at ijs.co.nz> Auckland, New Zealand
Javascript MEDLINE: http://www.ijs.co.nz/med/medline.htm
Ontario Ombudsman's Fairness principles:
http://www.ombudsman.on.ca/pdf/fairness_stds.pdf
More information about the Election-Methods
mailing list