[EM] "Approval" name out; VM17 Schulze theory fails; 80k Sincere Votes equals 80k Voters?
Craig Carey
research at ijs.co.nz
Sat Jan 29 00:29:00 PST 2005
CONTENTS
(1) STEVEN BRAMS APPROVAL IS INFERIOR TO MY BLOCK VOTE VARIANT
(2) LANPHIER'S "SINCERE PREFERECNCES": A CONCEPT SIMILAR TO HOW MAINFRAMES
· AND BSD AND VAX MACHINES ETC. CAN SWAP-OUT TO DISK, WHOLE PROCESSES
(3) MR VENKZE MAKES CLAIMS ABOUT THE SCHULZE METHOD
(4) SOME MINOR QUESTIONs FOR MR SCHULZE
As usual, this may contain missing words, letters, and missing commas.
It seems that people who copy from published articles, instead of thinking
out all the simplest cases first, can achieve leaps and bounds less.
(Dismounting can speed progress).
Today I discovered the ApprovalVoting mailing list, just after deciding to
develop a far better method for any in the Commonwealth that might need it.
What a surprise: I saw that Mr Marcus Schulze was writing at the
ApprovalVoting mailing list.
He was quoting Moulin's Participation, seemingly without having thought
about it in the 3 paper 1 winner 3 candidate case with papers (AB),(B),(C).
Maybe Ron Holzman shot it up before me. I can't recall.
By the way, maybe Schulze can start to aquaint himself with how this citable
Israeli voting article author spoke about the simplex, which is a type of
polytope. Mr Schulze seems to just never use the word "polytope". I should
archive the reasoning for that, provided less than 250 pages in length.
>Delivered-To: research- at _ijs.co.nz **2
>Date: Sun, 10 Sep 2000 14:01:35 +0300 (IDT)
>From: Ron Holzman <holzman- at _techunix.technion.ac.il>
>To: Craig Carey <research- at _ijs.co.nz>
>Subject: Re: Preferential voting, shadows of polytopes
>
>Dear Mr. Carey,
>
>Thank you for your message. Indeed I wrote a paper, published in Discrete
>Applied Mathematics 22 (1988/89) 133-141, in which a simplex point of
>view to voting methods was suggested. I have not been active in this area
>recently, and I am not aware of any other investigations in this spirit.
>
>Yours,
>
>Ron Holzman
http://www.math.technion.ac.il/people/holzman/index.html
This is going to the single-transferable-vote mailing list since finding
new problems with the decision of Mr Brian Wichmann (of England) who passed
the Monotonic Clones article of Mr Marcus Schulze. Voting Matters is a
research journal on the topic of STV and AV.
------------------------------------------------------------------------------
(1) STEVEN BRAMS APPROVAL NAME BURNS AND DIES
I nearly never write seriously on Approval. I drafted this message for my
single-transferable-vote mailing list. Then I chanced to discover the
ApprovalVoting mailing list, so I will send a copy out to that list.
http://groups.yahoo.com/group/ApprovalVoting/
Approval is not the best method yet it is being advocated to USA. USA might
get a worse method possibly because it has a better known name.
Here are websites of US registered Approval advocacy agencies:
http://www.approvalvoting.org/news.html
http://www.approvalvoting.com/news.html
Those websites are partly owned by the creator of the "Approval":
· ·Steven J. Brams, ·Department of Politics,· New York University,
· ·E-mail: steven.brams- at _nyu.edu
There is a 2003 article on Approval (and on how US politicians have been
rejecting Approval), by Mr Brams and Fishburn (of AT&T), here:
http://ideas.repec.org/p/cvs/starer/03-06.html
That mentions "negative voting" of Boehm, 1976: it's First Past the Post
except that a voter's vote can have a weight of +1 or -1.
An enveloping grander method I myself have devised, is the multiwinner
extrapolation, i.e. SNTV (fill 1 checkbox) where "voters" have "votes"
that have weights of +1 or -1. Presumably Mr Steven Bram's should cite
my name. Despite a core problem of IRV-ism and Approval-ism being the
copying of names, I suspect my name won't be copied.
-------------
· · · * The class of methods around the Approval of Brams and Fishburn *
As soon as the intelligent American hears about Approval then they would
rapidly construct a a family of methods that contains Approval.
Traits of the class can be:
· ·(a) There are nc candidates and nc checkboxes;
· ·(b) There are nw winners (or: nw seats);
· ·(c) The ballot papers shows some advice saying that the number of
· · · ·checkboxes to mark, is between p1 and p2. The advice is on the
· · · ·ballots and there is no OSSIPOFF-ish statement saying voters will
· · · ·behave sensibly (or else nothing).
Here are some "Block Vote variants", one of which is Approval:
· (1) The Block Vote method · · · : p1 = p2 = nw
· (2) Steven Bram's Approval· · · : p1 = 0, p2 = nc,· ·nw = 1
· (3) OSSIPOFF's inferior Approval: p1 = 0, p2 = nc-1, nw = 1
· (4) Craig Carey's Block Vote variant (try no. 1):
· · · · · · · · · · · · · · · · · · p1 = 1, p2 = Min [nw, nc-1], nw is free.
It looks like that my method is better.
Mr Bram is of a political department (Department of Politics, New York
Univrsity). Info: http://ideas.repec.org/e/pbr76.html
The Alternative Vote might get discredited before 10 years are out
possibly freeing up the "AV" acronym for the Approvalists.
------------------------------------------------------------------------------
(2) LANPHIER'S "SINCERE PREFERECNCES": A CONCEPT SIMILAR TO HOW MAINFRAMES
· AND BSD AND VAX MACHINES ETC. CAN SWAP-OUT TO DISK, WHOLE PROCESSES
Here I point out that the Election Methods converts ballot papers into
people by adding the word "sincere". To the STV community it would
all be a collection of ideas that should and would be discarded.
However members of Lanphier's mailing use belief and understanding.
Below is a bit of evidence showing that MIKE OSSIPOFF of the Election Methods
List expects that every voter will not tick all 100% of the checkboxes on an
Approval ballot paper
EML : http://lists.electorama.com/pipermail/election-methods-electorama.com/
No.1 ____________________________
http://groups.yahoo.com/group/election-methods-list/message/5436
>From:· "MIKE OSSIPOFF" <nkklrp at h...>
>Date:· Mon May 29, 2000· 6:52 am
>Subject: ·Re: [EM] Approval Vote: 99% isn't enough
>
>[Example of me (CC)]
>
>|· 99 ABCDEFGHIJ
>|· 1 J
>
>In Approval, no one has reason to vote for their last choice, and one always
>has reason not to. Likewise, obviously if you vote for everyone, the effect
>is the same as if you didn't vote. That's always contary to your interest if
>you have any preferences among the candidates, as your A voters do.
_________________________________
No.2 ____________________________
http://groups.yahoo.com/group/election-methods-list/message/5449
>From:· "MIKE OSSIPOFF" <nkklrp at h...>
>Date:· Thu Jun 1, 2000 ·1:37 pm
>Subject: ·RE: Approval Vote: The unfairness of being dead
>
>[Example of Demorep]
>
>|· 2 AJ
>|· 1 J
>
>In the above example, the A voters are voting for their last choice,
>something that no one would ever have any reason to do in Approval.
_________________________________
There is a problem there: OSSIPOFF seems to be believing that if he
drops in the word "sincere" and changes the word "preferences" into the
words "sincere preferences", then he can remove 200,000 (say) Approval
ballots, and replace that with 200,000 people.
Possibly over 90% would not say what their "sincere preferences" are,
since it seems that no one asks them to disclose their opinions.
The EML mailing list has other men who believe that "sincere preferences"
is an idea worth knowing about.
My e-mail records start in 1998 and it seems that Bart Ingles was using the
term ·"sincere preferences" in December 1998:
_________________________________
>Date: Sat, 19 Dec 1998 02:31:27 -0800
>From: Bart Ingles <bartman at netgate.net>
>X-Mailer: Mozilla 3.04 (Win95; I)
>To: election-methods-list at eskimo.com
>Subject: Cost of a vote
>Resent-From: election-methods-list at eskimo.com
>Reply-To: election-methods-list at eskimo.com
>X-Mailing-List: <election-methods-list at eskimo.com> archive/latest/2487
>X-Loop: election-methods-list at eskimo.com
>Resent-Sender: election-methods-list-request at eskimo.com
>X-Envelope-To: research at clear.net.nz
>
>Just now realized that there are apparently two Saari's.
>
>While I don't agree with Donald Saari's view of Borda being the best
>method, I agree with Mike Saari's view that in order to keep voter
>ratings somewhat in line with sincere preferences, a "cost" should be
>associated with a vote:
>
> http://www.egroups.com/list/election-methods-list/355.html
>
>I don't believe the cost should be monetary, however.· I prefer the cost
...
>B.I.
_________________________________
Here is another "sincere preferences" definition that relies upon the existence
of the "we" concept.
_________________________________
| Resent-Date: Thu, 12 Apr 2001 10:05:48 -0700 (PDT)
| Date: Thu, 12 Apr 2001 19:09:20 +0100
| From: Martin Harper <mcnh2- at -cam.ac.uk>
| Reply-To: martin- at -myreddice.co.uk
| Organization: Cambridge University
| To: election-methods-list at eskimo.com
| Subject: Re: I told Rob Richie, use "IRV", its catchy (was Re:[EM]Approval
| · · · · · · · · · · · · · · · · · · · · · · · Elections & Effective Weights
| Resent-From: election-methods-list at eskimo.com
| X-Mailing-List: <election-methods-list at eskimo.com> archive/latest/5828
|
|
| I was going to write a longer reply, but this seems to be the crux of the
| matter:
| ...
|
| 6) What's the definition of sincerity?
|
| My definition is:
|
| [subdefine] A greater insincere preference is a case where we prefer A to B,
| but express a preference on our ballot paper for B over A. A lesser
| insincere preference is a case where either we prefer A to B but do not
| express a preference, or we have no preference between them but we express
| one.
|
| [maindefine] A sincere vote is a vote which does not have any greater
| insincere preferences unless voting them is necessary to avoid other greater
| insincere pai[r]wise preferences. In addition, a sincere vote does not have
| any lesser insincere preferences unless voting them is necessary to avoid
| other lesser insincere pai[r]wise preferences
|
| Note that other people have other definitions - but I like mine. My
| definition of effective weight works with any definition of sincerity. Note
| that my defin[i]tion should not be applied to determine the sincerity of
| more complicated vote types than those used in Condorcet, Approval,
| Plurality such as those used in Dyadic Approval or Average Ratings.
| ....
_________________________________
The word "we" is creating an undefined weighted sum.
------------------------------------------------------------------------------
(3) MR VENKZE MAKES CLAIMS ABOUT THE SCHULZE METHOD
The main idea here is that a novel way to get answers out of believers in
Condorcet variants, is to ask the wrong people. It seems that they can
undo the problem of methods being fully undefined by using guessing or
soem other technique.
The Schulze article is here (Voting Matters Issue 17, 2003):
· ·http://www.mcdougall.org.uk/VM/MAIN.HTM
· ·"A New Monotonic and Clone-Independent Single-Winner Election Method"
I may have got this mystery of the Schulze method broken.
The topic is sure to be boring for some.
Here is the start of Mr Schulze article. Note:
The words "strictly prefer" are undefined. That is a major problem.
Also the method is missing from the paper that analyzes it.
Here is the start of the Schulze article:
| · · ·A New Monotonic and Clone-Independent Single-Winner
| · · ·Election Method
|
| · · ·Markus Schulze has studied mathematics and physics at the
| · · ·Technische Universit¨ at Berlin.
|
| · · ·In 1997, I proposed to a large number of people who are
| · · ·interested in mathematical aspects of election methods a
| · · ·new method that that the implementations had a runtime of
| · · ·...
|
| · · ·Today, this method is promoted e.g. by Diana Galletly
| · · ·[1], Mathew Goldstein [2], Jobst Heitzig [3], Raul
| · · ·Miller, Mike Ossipoff [4], Russ Paielli, Norman Petry,
| · · ·Manoj Srivastava, and Anthony Towns and it is analyzed
| · · ·e.g. in the websites of Blake Cretney [5], Steve Eppley
| · · ·[6], Eric Gorr [7], and Rob LeGrand [8]. Today, this
| · · ·method is taught e.g. by James E. Falk of George
| · · ·Washington University and Thomas K. Yan of Cornell
| · · ·University [9]. In January 2003, the board of Software in
| · · ·the Public Interest (SPI) adopted this method unanimously
| · · ·[10]. In June 2003, the DEBIAN Project adopted this
| · · ·method with 144 against 16 votes [11, 12]. Therefore, a
| · · ·more detailed motivation and explanation of the method is
| · · ·overdue.
|
| · · ·There has been some debate about an appropriate name for
| · · ·the method. Some people suggested names like “Beatpath
| · · ·Method”, “Beatpath Winner”, “Path Voting”, “Schwartz
| · · ·Sequential Dropping” (SSD) or “Cloneproof Schwartz
| · · ·Sequential Dropping” (CSSD or CpSSD). However, I prefer
| · · ·the name “Schulze method”, not because of academic
| · · ·arrogance, but because the other names do not refer to
| · · ·the method itself but to specific heuristics for
| · · ·implementing it,...
|
| · · ·It is presumed that each voter casts at least a partial
| · · ·ranking of all candidates. That means: It is presumed
| · · ·that for each voter V the relation “voter V strictly
| · · ·prefers candidate A to candidate B” is irreflexive,
| · · ·asymmetric, and transitive on the set of candidates. But
| · · ·it is not presumed that each voter casts a complete
| · · ·ranking. That means: It is not presumed that this
| · · ·relation is also linear.
|
| · · ·Suppose that d[X,Y] is the number of voters who strictly
| · · ·prefer candidate X to candidate Y. Then the Smith set is
| · · ·the smallest non-empty set of candidates with d[A,B] >
| · · ·d[B,A] for each candidate A of this set and each
| · · ·candidate B outside this set.
|
| · · ·5.1 Pareto
| · · ·Pareto says that when no voter strictly prefers candidate
| · · ·B to candidate A and at least one voter strictly prefers
| · · ·candidate A to candidate B then candidate B must not
| · · ·be elected.
| · · ·The Schulze method meets Pareto.
The words "strictly prefer" define a sum but leave these all undefined:
(1) ·the number of terms in the sum;
(2) ·the weights in the weighted sum
(3) ·the categories with subtotals that vote counts are added to.
Obviously fully undefined methods are not passed by Monotonicity.
---
The text of the Pareto rule shows that Mr Schulze gave a different
meaning to the same 2 words "strictly prefers": since no matter which
category of the 5 the paper is in, no weights are used.
For the first time I note that Mr Schulze seems to have misunderstood the
English language meaning of the words "strictly prefer" and the meaning was
not the meaning that he intended to have.
Here's my argument: Firstly, I assume that the words "strictly prefers" have
a meaning that is identical to the meaning of the word "over". I.e.
· (1) Candidate A is over B in the paper 1*(AB),
· (2) Candidate A is not over B in the paper 1*(A).
We can visualize it all: a black card is either over or under, a white card.
Presumably that interpretation will lead to a Condorcet variant that is not
what he wanted.
Also Pareto is a wrong rule for crashing on a 1 candidate election and also
for being wrogn in all 3 winner 3 candidate elections.· Mr Schulze knew that
Pareto was an unacceptably wrong rule. He might have been trying to copy from
economists.
---
The Venkze version of the Schulze method MUST BE rejected since failing
a rule that is a strengthened Monotonicity (i.e. Woodall's
Mono-Raise-Random rule).
· · · · · · · · · · · · · · · · ! ! !
Unexpectedly a breakthrough occurred when I got these lines ("For
3-candidate Schulze ...") from Mr K Venkze (who is in Minnesota, USA).
_________________________________
To: Mr Venkze
Date: Thu, 20 Jan 2005 16:32:49 +1300
| At 2005-01-19 08:06 +0100 Wednesday, Kevin Venzke wrote:
...
| >For 3-candidate Schulze (or Tideman, Heitzig, or Minmax, since
| >all are the same with three candidates), I believe this is
| >correct:
| >
| >A-wins :=
| >((a+ca)>(b+cb) and (a+ba)>(c+bc)) or
| >((a+ca)>(b+cb) and (b+ab)>(c+ac) and (c+bc)>(a+ba) and
| > · · · · · · · · · (c+bc)<(a+ca) and (c+bc)<(b+ab)) or
| >((a+ba)>(c+bc) and (c+ac)>(b+ab) and (b+cb)>(a+ca) and
| > · · · · · · · · · (b+cb)<(a+ba) and (b+cb)<(c+ac));
| >
|
| That simplifies to this (with infinitesimal mismatches):
| A-wins =
|
| ha2 := ( b + cb < a + ca
| · · ·and c + bc < a + ba)
| ·or ( · ·b + cb < a + ca
| · · ·and c + bc < a + ca
| · · ·and c + ac < b + ab
| · · ·and c + bc < b + ab)
| ·or ( · ·b + cb < a + ba
| · · ·and c + bc < a + ba
| · · ·and b + ab < c + ac
| · · ·and b + cb < c + ac);
|
_________________________________
It seems that Mr Venkze can arrive at the Schulze definition in one hop,
possibly by using guessing. I didn't realize that quickly arriving at
Mr Schulze intended meaning was so easy.
I simplified Mr Venkze's Schulze expression down to expression "ha2"
expression using the http://www.ijs.co.nz/ijs.co.nz/polytopes.htm
software. REDUCE for Win98 is at http://tope.tigris.org/
Now The Schulze method has this unacceptably wrong term in it:
· · (c + ac < b + ab).
The Alternative Vote has (c<b) face and it does not seem to be any worse.
However that Alternative Vote term is failed by Monotonicity and the
(c+ac < b+ab) term of the Venkze Schulze method would be passed.
The explanation is that Monotonicity is too weak.
This is how (c+ac < b+ab) it passes by Monotonicity:
· ·if a (B) paper is changed into an (AB) paper then candidate A is
· ·not harmed.
So possibly the Schulze method is monotonic.
However he made the same mistake as he did with Pareto: he copied from
economists. It is totally obvious that (c+ac < b+ab) is as bad as the
Alternative Vote's (c<b), so Monotonicity is too weak when Truncation
Resistance is absent.
That's becasuse A is harmed on adding -(B) + (A) papers.
So the barely-interested STV Community can reject the Venkze version of the
Schulze method.
Mr Schulze didn't mention tests that didn't pass his method.
The selection of tests can be finished before the methods to test, arrive.
--
Also, Mr Venkze probably IGNORED the whole PDF article of Mr Schulze since
his "A-wins" expression had only 5 lines. Here is my argument:
· ·The 3 "for loops" of the Schulze thing would lead to an expression that
· ·would be nearer 90 lines long.
· ·That polytope expression should simplify greatly, since it has to simplify
· ·into the "ha2" expression (unless ha2 is wrong).
To tighten that up, I would need to find another student of pairwise
comparing who can know what his colleagues mean, even when it is impossible
to get it from a careful analysis of their wording. However most don't
reply to me in a full way.
--
Also, to be logical, Mr Schulze's monotonicity proof must be discarded
because it was not checking the surfaces between the cases, that are
created by the indices and subscripts.
Here is an example showing a fault line:
A-wins· · · /
· · · ·P· ·/
· · · /|· /
· · ·/ | /
· · / ·|/ · A-loses
· ·/· ·Q
Between P and Q there is an infinitesimally boundary between 2 of Mr Schulze's
cases. It must be checked before it can be concldued that the slopes are
within the bounds.
These words suggesting shifting paths can be creating cases.
| · · ·For every candidate B other than candidate A the value p[A,B]
| · · ·can only increase but not decrease with d[A,X] - d[X,A] since
| · · ·only AX but not XA can be in the strongest path from
| · · ·candidate A to candidate B and the value p[B,A] can only
| · · ·decrease but not increase with d[A,X] - d[X,A] since only XA
| · · ·but not AX can be in the strongest path from candidate B to
| · · ·candidate A. Therefore ...
The proof seems to be faulty. If Mr Schulze had of preferred an
algebraic proof, it could have been true, but stuck in 3 candidates.
However he seems to have very very large possibly very simplifiable A-wins
expressions.
---
Anyway, to conclude, the Schulze (1-winner) method is rejected since it fails
Dr D. Woodall's Mono-Raise-Random. Ditto with Mr Heitzig's method etc.
The Alternative Vote is failed too. I guess the Alternative is better than
the Schulze method.
Mono-Raise-Random shall be called Monotonicity-2 by me.
Here's the idea of Monotonicity-2:
· ·Candidate A is not harmed by moving the "A" preference one preference to the
· ·left (towards the 1st, with non-negative weights).
· ·Also all types of rewriting to the right of the "A" preference are allowed.
---
A better looking fairer Condorcet method than the Schulze method, might be
Mr Venkze's MMPO (1-winner) Condorcet variant, since seeming to not fail
Monotonicity-2 when 3 candidates.
I have not seen public e-mail communications of on an MMPO Condorcet variant
so I won't write on it.
I started on Monotonicity-2_And_Etc rules at the
http://groups.yahoo.com/group/politicians-and-polytopes mailing list.
------------------------------------------------------------------------------
(4) SOME MINOR QUESTIONs FOR MR SCHULZE
[1] Can Mr Schulze's Condorcet variants get past a rule that says that SNTV is
· ·embedded ?.
· ·I assume you can repsond "yes". In fact let's assume that you will reply
· ·with a "yes" (using the Venkze breakthrough discovery: simply explain
· ·what Mr Schulze have been thinking without bothering to consult).
[2] Suppose there are 2 candidates, 0 winners, and these are the papers:
· · · ·2 (A)
· · · ·1 (B)
· ·After resolving the Schulzian beatpaths, the Schulze Condorcet idealism
· ·would still find 1 winner, and the "Right Number of Winners" rule says that
· ·the number of winners is 0.
· ·I guess that Mr Schulze would want to drop a (ideally unfollowable)
· ·reference to a dud published article of an economist.
I can attempt to map put the size of the epochs of progress that Mr SChulze
will be making:
(1) Mr Schulze has been at the Election Methods List for 2,556 days
· · ·(7 years).
(2) The complextity of Marcus Schulze polytopes rises fast, perhaps
· · ·in proportion to (n!)**5 where n is the number of candidates.
(3) Since Mr Schulze can't solve all 0 winner elections, and hence can't
· · ·solve the the 0 winner 1 candidate election, I copnclude that Mr Schulze
· · ·can't figure out an idealism for all 1 candidate elections.
· · ·Thus Mr Schulze's competence starts and stops with this 0 candidate
· · ·election:
· · · ·z*()
(4) I get the result that Mr Schulze will have be competent in 4 candidate
· · voting maths in 55.74 million years (PS. binary star systems are frequent:
· · We'd just divert a sun and if Earth can enter a galactic community and
· · plead effectively). [Aside to Forest Simmons., will you draft up a speech
· · for MIKE OSSIPOFF Fruticake, the sincere preference man ?.]
· · However mastery of 3 candidate idealism could be achieved by Mr Schulze,
· · in about the 54,432-th year.
· · I don't get read much from Mr Robert Lanphier: thoughts, ideas, facts,
· · analyses, e.g. a 48 point plan for stopping it from being dominated by
· · Americans and Germans aspiring to give wrong answers or none.
My 20MB REDUCE for Win98:
http://tope.tigris.org/servlets/ProjectDocumentList?folderID=4137
Find-A-Drug : http://www.find-a-drug.com/ Matching drugs against models of
proteins.
At the moment most points are awarded for "Methodology Research". It
changes about every month).
perl -S adddots.pl < out.txt >| out2.txt
__________________________________________________________________________
G. A. Craig Carey, Auckland, New Zealand
(11 candidate Twin Towers Cascading Nobodies defect of the
Alternative Vote: http://www.ijs.co.nz/irv-wrong-winners.htm )
Russian Venera spacecraft: http://www.astronautix.com/project/venera.htm
Nth pole not landed on
More information about the Election-Methods
mailing list