[EM] Approval Vote: The unfairness of being dead

Craig Carey research at ijs.co.nz
Tue May 30 14:24:04 PDT 2000

[version 2, GMT 30 May 00]

I have rejoined the mailing list. I quit a while ago and sent to
my list, messages demonstrating the shadowing of polytopes. Then
I got the 6 papers 1st subproblem of the 1 winner 4 candidate
problem shadowed to a conclusion where different a new technique
is needed to progress.

I am returning to this list to help out in the battle against the
evil that it still has it has not repelled -- the advocacy of the
Approval Vote using imprecision and lack of principle.

I much presume that councils are not interested in the Approval
Vote. They probably want good methods.

The EM List,

The Politicians and Polytopes list


(The 2nd to last message at the EM List's Egroups archive is
message number 5440, Tue May 30, 2000 12:16pm)



At 10:12 30.05.00 +1000, David Catchpole wrote:
 >On Tue, 30 May 2000, Craig Carey wrote:
 >> Thanks David. It makes us all seem so thoughtless. I specifically
 >> include you now that you said you are not impressed.
 >What motivated my nasty message is the simple fact that you have indicated
 >a positive feature demonstrated by approval voting and most (if not
 >basically all) methods- Pareto efficiency. In the case you give, 99% of
 >the population are indifferent between all the candidates. That the other
 >1% does have an opinion and expresses a preference for J means that the
 >only preference expressed is that for J. For the election to be Pareto
 >efficient, J must therefore win. It's quite defensible (and no mention of
This Pareto idea is either undefined or its dumb. Mr Cretney's page got the
definition too wrong for me to have reconstructed the intended definition,
and I have not searched elsewhere on the Internet. I could be about to
discover a really bad idea and I wonder if David could say more. (Sorry,
I am not as learned as many other).

What David (and Mr Ossipoff) are getting close to saying, is that
preferences don't indicate the paper's preferences, in the Approval
Vote. If that were fully true for the Approval Vote, it would be
much better named "Limited Voting", which is defined on


except that the number of votes may exceed the number of candidates,
except that quite probably the ballot papers would not allow that.

To fail to fully disregard preferences leaves the accusation that
you have not got the Limited Vote but instead something that respects
preferences. Tell me again: there are no preferences in the Approval
Vote?. Then there never was an Approval Vote. Perhaps when you again
began writing to list, you gave it the wrong name. That sort of thing
would not happen in the UK.

Mr Ossipoff wrote this and it is false: "99% of the population
are indifferent between all the candidates". Mr Ossipoff never said
how he arrives at these figures.
I state Mr Ossipoff does not even have a probability.
So everybody has to ask: has Mr Mike Ossipoff constructed a new theory
of relational comparisons that handles that compares given numbers
(probabilities) against fully missing observed numbers.


I shall quote what I wrote on 30May00 to a person that was not
satisfied with the government committee's rejection of STV electorates
and their recommendation of a regional, not national top up scheme:

 >Lord Alexander wrote about shipping: if an engineer was learning

  [UK shipping is in trouble, with distorting influences on investment
  and so on, and the Lord (constrained by government) turned his mind
  to that.]

 >about the stress that ships could endure, he would model the
 >metal, and then consider only the model, not talk about ships
 >[voters' wishes] when the model was in trouble, and the model
 >when the ships were in trouble [no problem, their methods are
 >ignored]. That may be enough to get rid of public interest
 >and public good, in the theory of counting votes (and to defend
 >a little more, Ms T...

I will completely ignore the analogies to the ship sinking and
keep the emphasis more how a failing model is defended with reality
(rather speciously referred to by Mike O), and yet in the background
it seems that mathematics gets sunk moments after opponents move in
demanding increased precision lest the assertions that seemed
designed for the lesser task of misleading, are proven to be false.
A ship can be lost, but the sky and sea as well, is enough to
restart the possibility of a disagreement between me and the author.

I know the rules: it is a zombie and if even if pumped full of
bullets, any one of which would kill an attentive dog, is not going
to remove the Approval Voting method. This is one of those brown
cockroaches I wrote of last year. The servant of the politicians,


 >> Until I had that message sent in, it might have been a method
 >> for Australia, unless the Australians can identify bad methods
 >> when they see them. Those EM people need to get a questionaire.
 >> Can you do that for me David?.

The list could have a common set of strict principles?, and have
occasionally, an idea added and less often, a common idea lost?.
A big problem with the list is that people can't tell what it is
that they lose for as long as they temporarily accepts someone
else's ideas. However it is starting to appear the list had no
ideas in common and its handling of Mr Ossipoff's ideas suggests
that those who post messages have to be told things like how it is
that methods ought be 'fair' globally, which means that the
method has to sum votes correctly.

In this list, when someone temporarily asserts their belief, other
beliefs may have to give way. But no one seems to find out what is


 >From: MIKE OSSIPOFF  <nkklrp at h...>
 >Date: Mon May 29, 2000 10:54am
 >Subject: Re: [EM] Approval Vote: 99% isn't enough
 >Demorep wrote:
 >>research at i... asked me to forward yet another Approval example.
 >>to him and not me.
 >But I hope you don't mind if I reply to the list, because that's
 >where you sent Craig's letter.
 >>research at i...----
 >>Here is an improved example:
 >>   99   ABCDEFGHIJ
 >>    1   J
 >So those are the sincere rankings of those 99 voters, and that 1
 >>Total Votes = 100
 >>Number of Candidates = 10
 >>Approval Vote internal counts: A:(B..I):J = 99:99:100
 >I don't know that that line means, or what an internal count is.
 >I assume you're saying that 99 people voted for A, 99 people voted for
 >B through I, and that 100 people voted for J, meaning that all the A
 >voters voted for J, and the J voter voted only for J.
 >Excellent! I love these examples where the foolish voting is

There are no people around to accuse of foolishness. Mr Ossipoff would
put 1 million people on top of election room counting tables so that
they could be questioned, EXCEPT that here, as in the stating of the
probability, Mike doesn't collect data or ever suggest how to find out
what voters thought. Where does the money to pay the interrogators, come
from Mike?. How else can you find out that the voting is foolish?.
With such falsity, the verb "love" and adverb clause "particularly
extreme" can be found to not modifying or amplifying, meaning.

 >particularly extreme. That line that I didn't understand seems to be
 >saying that J got 100 votes. There are 100 voters. Everyone voted for
 >J. That means that all the A voters voted for J, their last choice. And
 >if B through I all got 99 votes too, then the A voters voted for
 >everyone. In Approval, no one has reason to vote for their last choice,
 >and one always has reason not to.

A method that behaves wrongly in a particular region might as well be
examined there.
How does that defend the Approval Vote from the proof that it is
unfair?: to accuse voters of not having a reason. It is no defence
against a instance of unfairness, to probably wrongly tell voters what
they could have done, when the issue is the selecting of the wrong winner
by the method.

It is clear from the last sentence there of Mike Ossipoff, that he regards
the Approval Vote to be different from the Limited Vote. In the Limited
Vote, there is no last choice. [I don't see any final way to defend the
Approval Vote except to say it was misnamed and all along he was writing
about the Limited Vote (or a version of it).]

Just to be sure: do people use tick/crosses on Approval papers, or numbers
(!, 2, 3, ...) ?.

 >Likewise, obviously if you vote for everyone, the effect is the same as
 >if you didn't vote. That's always contrary to your interest if you have
 >any preferences among the candidates, as your A voters do.
 >>Approval Vote winner: J
 >>Candidate A has 99% of the first preference votes and loses
 >>Candidate J has 1% of the first preference votes and wins
 >Mike Ossipoff

The world is my toilet


This message is about Mr Ossipoff's finding that

   1 J

was not a significant example in his opinion.

He stated it was improbable. I note that Mr Ossipoff is becoming a
frequent exponent of putting on a seemingly false ideas that no one
can take away from him and indendently check. It is as if
conclusions of a theorem for which there is no attempt at a proof,
can't be checked, not because Mr Ossipoff admits the proof doesn't
exist in checkable form and the opinion cannot be proven true, but
because data or ideas in the proof [still not seen] are trademarked
or copyrighted and secrecy prevails.

Perhaps Mike will agree he has no data to reject the example. However
we perhaps will disagree over whether it is valid to come to the
conclusion withouit premises that make a basis for it.

Specifically, I wrote about probability. If a statistician read the
argument, he could ask for the distribution or data. What seems
amazing to me that is that I have discussed this exact matter in
private correspondence. Surely it is now an uncorrectible ingrained
habit that the list must endure, to wheel out the same dead
statistical argument/body every time his own ... theory gets even
deader still?.

This is a technique of claiming improbable is now known. Can Mr Ossipoff
show it is not just an attempt to mislead something. What is mislead?:
I am reluctant to name the list, yet it is tricky to think of something
else. I don't see why Mr Ossipoff would want to merge the the world
and mathematics. Surely it sunk with paradox if it was the aim to retain
control in both worlds.

I request, under the guidelines of the list, the probability of this
  example ever occurring:

    1 J

"If the honourable member says he has the information then our party will
allow sufficient time for that information to be collected and provided.
How long will it take the civil service to get your information collated
and presented, Mike?".

I seem to recall having trouble with this type of question before.
So I will clarify my purpose. I want no misunderstanding: this is not a
request for evasive response being made, with it almost justified by me
having an ulterior motive. Instead pure truth is my aim and I just want
and request the probability of the example. It is nothing a request for
"rope" and that a mathematician unhampered with irrelevancies would be
at all concerned about.

This letter is not something Mr Ossipoff has not expected: I wrote to
him saying that a defence using false statements was not what I wanted.
I asked for fallacies to be presented withot muddying English, but that
was not what happened.



At 02:46 28.05.00 +1200, Craig Carey wrote:
 >A conclusion would be that the Approval Voting method is a method that
 >very little fairness for some examples. My list prefers to exclude
 >worst case examples whereas at the EM list, they like to try remove all
 >the problems (at times) in a rough sweep.

Is the Approval Vote a method that has continuous surfaces and that has
bounds on the normal vectors of surfaces of win-regions?. Of course it
is. Then it is true, that it is possible that global properties can be
assured compliant with some rule, only if extreme cases are considered.

Let me say what I think are the details of Mr Ossipoff's thinking. The
touching of tangent cones to propagate wrong win outcomes is a simple
idea and I have just stated it. It is a possible way that permits
overall statements [i.e. throughout the simplex with papers at vertices]
to made. Mr Ossipoff said that the presence of an extreme case had to be
dismissed since improbable. The probability of all election examples
would be infinitesimal if real numbers are used. I ask that the
implicit range or integralness assumption be made explicit.
Mr Catchpole missed that. What is the volume of a point?.

To find out if Mr Ossipoff never ever held an estimate of the
probabilities, I ask: What is the estimate of the probability of an
instance of any candidate of the Russian Duma in 2000-2021 in the
circumstance they switch to the Approval Vote, losing with 50-100% of
the vote?.

If the data was lost, then can you find the figure?. If it really was
lost, what is your best recollection of the probability of the example?.

Mr Ossipoff and I permanently disagree over whether real data from any
time or civilisation is admissible. I always believe I lose; perhaps the
very existence of voters [and perhaps even public interest] creates
a complex universe of possibilities for defending extremely bad methods.
But it also the case that opponents can leave the real of mathematics
to enter the fantasy of personal voting theocracies, and criticise.
I would welcome Mike saying that henceforth he has stopped cutting up
ladies legs to insert them into his formula in a netherworld where
fairness was lost and voters replicate opinions held for them by the
method defender, in the undesirable millions.

Note it is a sound trick for me to reduce the 99% figure I gave, and
take it back down to 50%. I am sure expects he was set up, but the defence
will continue. Rest assured Demorep, the Approval Vote is far from a
rejected idea so far. We haven't yet got to where the adversary says he
  won't read what I write (or read and write).

Correct me please if I have the algorithm wrong:

This is how the Approval Vote can be defended from implied badness from
extreme examples. Perhaps Mike can argue that this is so extreme that the
reasoning used for saving the method, couldn't have significantly existed.

 >  For Approval and other methods like STV, two statistics are calculated:
 >  t[i] = (Badness of method i in the example [with some new units, IPVSBU
 >           [the international preferential voting standardised badness unit])
 >          *
 >         (Probability of example i [estimated without data or derivation;
 >           although in the UK they actually use data, when predicting
 >           defect incident probabilities with the Alternative Vote])
 >          *
 >         (Probability that over both cases this entire argument has not
 >          been rejected by the thinking consensus of the attending
 >          mathematicians)
 >The reasoning proceeds to the test: is t1 > t2 ?.

(The 3rd term only has an effect iff it equals 0, of course.)

Can the list clarify for me the acceptability of the use of that argument?.
If it is good now then it is good to use in the next 3 years, surely.

Can Mike tell the list what is the badness of my example
      1 J
Obviously Mike does not use the units IPVBSU.


I just sent a message to my list commenting on Blake Cretney's webpage
on rules. He defined the Majority Criteria as saying a candidate should
get above 50%. That is false, and this example confirms that:

Election system V =

     1: A

Number of winners = 0

People now have strong mathematical justification for ignore messages from
Mr Schulze (and Mr Eppley is it?). But the aim of the message is more to
give Demorep a defence. We can all imagine how he feels through the events
where a clean and thorough win using only a numerical example was done.

But the List is so rusty on voting theory, that I need to state here plainly
that the numerical example (99:ABCDEFGHIJ, 1:J) demonstrated 'unfairness'.
This is a bit of what happened:

  David Catchpole : unimpressed and not defending fairness. The others will
               fix. Take a seat.

  Mike Ossipoff : first hand knowledge of voter's understandings and beyond
               that, hard numerical data on probabilities obtaine perhaps from
               spirits and deities that know of the lord's 'O-tope', all 
               elections [rather than all possible papers, as happens in
               my IFPP voting theory]
                  Truth advisor to councils: B. Inglis.

  Mr Shulze : opposed to bad reasoning or not?. [Probably not.] Still, I may
               have written off Schulze theory. A side effect of getting an
               idea of Norm Petry written off. It is a significant event
               because it took 23 papers to get Condorcet, and that beats
               the 21 it took to write STV/AV off forever. I guess that means
               the Condorcet ideas might be a bit better than the Alternative
               Vote. However the details would need to be checked out:
               No need to reply, Markus. I know you no longer reply to my
               personal e-mail [due to the detail and length of the criticism]

  The backbone collective of the list: the archiving hard disks that are
                     subscribed: No comments so far.

  The Green Cerebrnull-Aliens repelled was by Maj General Ross Perot and the
               stat data collecting Oss-spirits defending the right to win
               the Presidency under a new system with a mere 0.0000014% of
               the vote by first preference counts.

I think Mr Ossipoff stopped advocating FBC after I had criticised it.
So it might be possible to suppress the Approval Pox that could be carried
from city to city. I ask Demorep: do you think I could improve the example
still, but after the geniuses in the advocacy group have come up with a
new false defence for the Approval Vote?. Maybe get the "2% A MO" down to
"2 A MO"

Mr Shulze wrote that having a method based on axioms is a test that
passes any method. No that is not something that you can do, and these
axioms precisely specify surfaces, rather than specify bounds for the
slopes of surfaces which is what would happen on attempting to reconstruct
axioms. Reconstructing axioms from existing methods may be difficult, and
the axioms could be similar complexity to the method. A method can be
briefly stated by giving its axioms. To contrast it sharply with Condorcet
variants, a method that has an axiom that it is maximally proportional,
is maximally proportional. This is the sort of reasoning that gets used
at my mailing list:


Mr Shulze doesn't seem to comment against the use of invalid reasoning.
I suppose this is irrelvant.

Also, the Approval Vote can be grossly unfair. The advocates of it know
that presumably. I don't regard that a back-scratching collective of
invertebrates (or is it politeness-advocates that are so unable to reject
bad rules/methods that they can not even reject my good ones), is a proper
place for the consideration and rejection of specious ideas and some
principles, e.g. fairness. That is lost by the Approval Vote.

Suppose the planet were devastated and harmed and Mr Rob Lanphier
himself was tracked down with FBI CALEA TCP/IP monitoring systems and
then was gruesomely crushed under a UFO?, emitting sounds [he writes to
other lists on IP sound protocols]. The United States president, Mr
Perot, drove off the alien green Ossibrainsloths because they lied to each
other and had a weak leadership elected by their Approval'Inglis B'Voting
system). In the aftermath of the war, the president (named 'A' here)
can and would lose against no-hoper candidate 'MO'. That led into the
second nuclear skirmish. Candidate A had 95% of the vote.

    4  A
   90  A MO
    6  MO

The Approval Vote elected Candidate MO, mainly because MO promised special
payments to those that voted for him with their 2nd preference.

Mike tells use that that is improbable. Tell me MO: if it were written
in a fiction book and the author of the book exactly stated that the
probability was one, then does every reader of the list need to restore
the Ossipoff deities or spirit collective, the ones that collect your
data prior to you sending it in to the list, or is it impossible for
the book to be written. I.e.the fiction book that gave the example above.

Let's just assume that fairness is regarded by the fiction author and
people in the story [exclude USA today], as being important. Inside
the EM List, fairness is to be sacrificed, apparently.

Computer programmers need to get their code bug free even if not sure
the code being tested will be called. Mike Ossipoff tells us he is
certain the Approval Vote is bad [cf with programmers fixing their code],
and he says the chance of it being used is definitely above 0. The
fix is doubtless to make the method more complex. However this is one
of the worlds simplest methods. Is there some principle we have not
been told about that prevents it being improved?.


Readers might be surprised by my comments, but they could note that I
believe that this is just the first part of Mike's Ossipoff's entire
pseudo-mathematical theory of defending absolutely bad and favored

Referring to the British ship modelling example again, and to Lord
Alexander, this style of reasoning that the list will hardly criticise
with any clarity (save for me), is analogous to an instance where the
bad mathematical model of the ship's steel and hull is saved (by the
engineer) by some quick distractions of a doubtful nature, and
references to the unquestionable reality of the ship. Otherwise the
engineer would fear a very bad problem happening to the theory.
The theory might be sunk at worst, or else criticised on purely
mathematical grounds (or those of quality of reasoning. Those that
may actually meet Mr Inglis or Mr Ossipoff can take up the other part
of the engineers dilemma, the part where the entire ship sinks. That
is perhaps just why Mr Ossipoff has been sending rules to this mailing
list. The fix is simple: separate mathematics from reality and get the
modelling right. Get rid of the beliefs that lead us to conclude that MO
is surrounded by omnipresent spirit helpers (or else are making overly
strong statements that are not true).

I haven't any doubt that Approval Voting will be bad for a public
election. It is not truncation resistant. It is not a fair method
locally (power of votes can vary by large factors, and effects of votes
can be negated since it is not monotonic [Cretney's page has that wrong
hasn't it?]), and it is not fair globally (the example just above), and
it is not truncation resistant. Further it seems to take a lot more
bullets through the brain before its slows down and stops being freely
advocated in open free public discussion forum

Craig Carey, Auckland. New Zealand 

More information about the Election-Methods mailing list