Re: Robonson wins [...]
At 03\04\22 13:10 -0500 Tuesday, Branden Robinson wrote:
> Re Robonson wins [4...].ems<file://v:\mail_archive\attach\Re Robonson wins [4...].ems <0880.0002>>
>*** PGP Signature Status: unknown
>*** Signer: Unknown, Key ID x2B46A27C
>*** Signed: 03\04\23 6:10:30 AM
>*** Verified: 03\04\23 2:37:21 PM
>*** BEGIN PGP VERIFIED MESSAGE ***
>
>On Mon, Apr 21, 2003 at 01:44:38PM +0200, Matthias Urlichs wrote:
>> In other words, if you held a vote which would ask whether to annul
>> the vote and replace Martin with Brandon, the majority would be
>> against that proposal.
--------------------------
At 03\04\21 13:44 +0200 Monday, Matthias Urlichs wrote:
>Summary of my point:
>
>This election has demonstrated quite nicely that those Debian developers
>who voted prefer Martin to any other single candidate. In other words, if
>you held a vote which would ask whether to annul the vote and replace
>Martin with Brandon, the majority would be against that proposal. I do not
>understand how you can, given that fact, declare that Condorcet voting is
>fundamentally flawed.
--------------------------
A problem is that at the end of competent arguing no competent in the
topic of selecting the correct winner could believe that the Condorcet
winner is the desirable or right winner. Mr Urlichs can't seem to name
the quantifier logic style axiom the guarantees that it is improper for
for one person to have too much power. Mr Urlichs was trying to say that
this Debian-vote should keep the whole [wrong] voting polytope and not
un-nicely discard it and with a sad show of unreasonableness, no one is
allowed to check that polytope at any other spot than that single
point. So far that sounds very unsatisfactory by the offered ideal of
nearly total ignorance might still allow rejection of the Debian voting
polytopes since it did actually get the wrong winner there. I can
complete 2-50 pages of algebra and derive the fair solution using
universally acceptable fair axioms, etc. No sign of disagreement so
far against the idea that votes should not be negated. Condorcet
allways negates votes on the surfaces and the election point of the
last election was close to a surface bounding the paradox regions and
it very likely a surface that was non-monotonic.
Happily Mr Urlichs reminds that pairwise beliefs are our favourites.
A nice social analogy is that malevolent hooligans who applied
excesive force and trickery in resolving disputes at each election,
could be in a cage for all their ability to produce wrong solutions.
A mathematical definition of equal suffrage would say something like
this:
"no one else's ballot has too much apparent power (especially as
preferences are added), and this ballot paper does not harm that
interests that it declares when in amongst the collection of
ballot papers."
That principle has a sharp definition: it is forever inconsistent
with the Condorcet idea. An obvious possibility is to put in a
fair method that has the last election be a fair 3 candidate
1/3 quota election that is monotonic.
Let's consider this: what if we got this explanation: there was
2 or more elections and the last was perfectly monotonic and it
had 3 candidates and had 2 stages:
* every candidate with less than 1/3 of the vote is rejected
with Alt Vote style eliminating;
* the winner is the winner of the remaining 1-2 candidate
election.
Presumably they won't feel like writing well on that. Suppose
they were perfectly blocked from making adjustments and then
were asked to adjust the algorithm to make it more like
Condorcet. I guess 100% of readers understand though Debian's
Condorcet likers can sustain unprincipled threads seeming to
discuss the methods by which unfairness can be advanced and
reshaped, their boat of their whole purpose would overturn if
let no option but to tweak-up a fair method into a bad one,
instead of their wrong dumb current plan of tweaking-up a
bad method into a good one. The future victims of the tweakers
are reading this. This is now two mailing lists. Of course
they slipped away. Check every single message of the
Debian Condorcet liker and see if there is any plain or
implied recognition that Condorcet is perfectly incompatible
with fairness. A fair algorithm might not get the winners
that the group desires, but it is about the only way to go
under an aim of avoiding sucking in the world's most unconving
bad-advice-giving Condorcet-likers or theorists. Now that a
few of them are subscribed: the big plan would be to keep the
rows and rows of very large machines that are true
mathematical off the top of their dirty green air balloon of
lies about what is good. All Martin has to do is tell them
that some actual mathematical quality claims on what makes
Condorcet better must be provided or else the big huge side
even of tweaking up the method is blocked and shut. Then the
Condorcet likers (who can do without justice themselves and
pure reasoning too) would vote against the current leader
who might then want an incumbent protecting methods. One
of the very options in that direction is a perfectly monotonic
method that has a very big 1/3 quota. I was not able to follow
it all but a plan to put a small quota in was all part an
parcel of the aging culture here of trying have a good
chance of being unfair to the previous winner.
Since Condorcet is identical to the Alternative Vote when
there are 3 candidates, then this could be considered:
some patched Condorcet method reduces the number of
candidates to three. Currently I solved the 3 candidate
voting problem on this page http://www.ijs.co.nz/quota-13.htm
Fairness is very agreeable particularly for persons who
do not like other making wrong adverse decisions about them.
Wrong adverse decisions is just the sort of thing that
Condorcet would like to give. The Condorcet likers might
want to pull away and "switch off". Check carefully and see
if they are on a winning streak in not knowing if I am
writing about the principle of keeping the power of a paper
in the range 0...1.
Debian has no competence in preferential voting. Last month
it could not get the winners right. This month its awfully
incompetent theorists are thinking of dropping out of the
Debian-votes list and creating a new internal mailing list
inside here. Those who did not make it out to the new
mailing but who got delayed here might want to pack up
their stuff and troop out to the segregated threads of those
want to vanish into a secret chamber, alter the Debian
voting system. A possibility is that Debian is not the only
Linux distro that has to keep renumbering its debian-vote
mailing lists each time the compent theorists moves into
communicate with the Debian Condorcet-liker. The Condorcet
enthusiasts (or apathetists) shipped themselves from Rome
to Greece. They never offered a principle that is a core
idea of the winning faction: the javelin that could prevent
ballot papers racing ahead in the march to the capital --
the rule to constrain the power of others.
The power to vanish into a secret chamber and thereby affect
the next winner.
In the following I pose some questions by way of proposing
some solution. One obvious solution is to shut down the
tweaking. Their seems to be a morally weak aim of increasing
Mr M's chances at the next election. I guess Mr Robinson
prefer my view of a much stronger defence for the current,
i.e. having the first step in the 3 candidate Alternative
Vote use a 1/3 quota instead of eliminating just the
candidate with the smallest (not that that 1/3 does not
extrapolate pleasingly to use of 1/k). That would: (a) confront
the lying that is Condorcet and (b) reduce change in the
winners. My previous proposal could alter that to whichever
candidate is quickest at delivering results during last-minute
debates. That could possibly lead to faster evolution at the
Debian system (and given what Condorcet likers think of,
Debian could move slowly if it somehow there is a
transferrence of character).
Later I might say how year of best effort improvements
could result in improvements in monotonicity if there was
not clear understanding on what monotonicity was. I observed
a tiny <1% improvement when altering an STV algorithm.
The absence of success in improving the system can be much
bigger problem than the use of wrong tests, lack of
knowledge on what to do. and complete absence of any checking.
If Mr V's pro-Debian webpage guides then there chould also be
false claims that sacrifices are need, checking is occuring
and statements on improvements that mismatch with what
happened. If Mr Urlichs lays down his thinking then I suppose
the new idea of running a Monte-Carlo program at 1 point per
year would be added. But I found that it takes about 3000
samples with 3 candidates and if publishable results are
needed then Monte Carlo and a 1GHz PC and 1 month is not
good enough for 4 candidate elections. However Debian produces
OS/390 software.
What would be nice is for IBM to supply a nonlinear optimizer
that actually runs well in the presence of perfectly Boolean
barriers. I searched and found nothing. I may have to write
one. If Mr Voss was actually intending to check the Debian
system then Debian might have a just cause for using that
currently-not-found nonlinear optimizer too.
High generals principals drop into Debian
Let the worst Debian has reduce the candidates to 3 and then
use the best method known to man in the last stage.
Case 1: If the last stage is the Condorcet method that got
patched up to make it as monotonic as possible subject to the
constraint that the Condorcet winner wins, then surely
that method is PERFECTLY IDENTICAL TO THE "IRV" METHOD.
("IRV" is an acronym of the Rob Richie think tank and they
renamed the Alternative Vote that is a standard English term).
The best 3 Condorcet variant for the papers (AB),(B),(C) is
the Alternative Vote, almost. It sort of requires an added
aim (i.e. keeping B losing when B has less votes than A).
The Condorcet winner and the Alternative Vote winner were the
same in the last election. So long as there is not a strong
4th candidate then Debian is actually using the IRV but
internally calling it the Condorcet method. It sounds like
some sort of internal self-deceptions. The advocacy of
the Alternative Vote centres around having 1-2 polytope
sequences and then pronouncing that only the Alternative
Vote is best. It sounds like the same method Debian uses
internally.
Why not send a communication off to the CVD and explain
to them the dilemma:
* given that the disliked monotonicity principle is a
way to convert Condorcet into the IRV method, does the
CVD advise paying some respect to the principle of
being fair to the contestants and their supporters?.
I suggest that persons join the "instantrunoff" mailing
list. It is quite an awful list and it is on the topic of
spreading the method.
The whole backyard of the meowing cats of Debian's
Condorcet likers could be photographed with the question:
why not simply use this method:
* the method is Condorcet and all the unknown regions are
shifted according to the desire of monotonicity and never
beyond the positions of the boundariess that the
Alternative Vote would place.
All polygons are polytopes.
The Debian project might have been subjected to a stealthy
attack from the fairvote.org people who oppose the fairness
in all known forms. In California the CVD had members of one
of its group advised by leader Steve Chessing, to join up
with another group. Then the 2nd group's votes would be
controlled.
Maybe a top leader of the 2nd group sabotaged its interests
which previously were unfriendly to IRV. To favour the
infiltrators with a 2nd preference through unawareness
could cause more to arrive. Wherever the CVD goes it offers
unbelievably weak arguments presuming nothing is checked
except what they say can be checked.
Condorcet likers are cultivating a style that hass nothing
but misfortune for Debian and that is having no answers and
no presence in any debate. No more than any leader can I
counter all those minute details on adjusting the Debian
voting system. It seems to be over motives that are not
true. I guess there is no use at all of a thought that would
ensure that the winner is right. The pairwise comparing idea
is an ancient idea that 21st century man can solve much better
by splitting it out into its 3 principles:
(1) truncation resistance (it is never upsetting to add another
preference)
(2) monotonicity. An attempt to help never ever harms
(3) _subject_ to the 2 above being held, "proportionality"
(i.e. for every preference, the weight of a paper naming a
candidate is added to its total).
That is about right.
Can someone assist me and do this for me.
Persons should have been told that the Debian project was being
IRV-ized.
There would be a spreading evil of IRVs unfairness without some
central intelligence ?. They would like to banish me but they
seem to be really quite fact averse. So they got themselves
banished. Many times I have seem the very same behaviour at
Rob Lanphier's own Election Methods List mailing list.
If I write down a formula that finds the winner then all
the faces are flat (not curved) and of each flat this can
be made available:
* the principle agreeable to rules providing rights that
set the orientation of the flat; and
* an argument with however many steps that shows why that
flat is in that position. The 1/3 quota follows from 2 steps
where 2 shadows are cast (If A wins then (B)->(C) has B stay
losing so C wins in the wedge; then (AB)->(C) creates the
1/3 shadow.)
There is something that smells quite suspect with the activity of
Manoj Srivastava.
I'd say that the very worst that the current leader can do is to
permit Manoj Srivastava to get what he wants.
Having so very many lightly untrue statements at the website of
Mr Voss, and having Mr Voss speak of being party to the activities
of Manoj Srivastava, implies to me that motives are shockingly
bad (and to that I can add, unfixable too). These people can't
argue in public -- well, at least if the topic is over their
core area of expertise -- and it would be fascinating to read
some of the private communications defending Condorcet at
Debian. Some of the arguments defending the retention of
Condorcet and its complete lack of principles would be arguments
that are also very clearly anti-Debian.
The presentation of voting methods in legal format is done with
a purpose including that of concealing that the method is not
monotonic. There is NOT any other prime aim (when 1 winner and
3 candidates) for Mr Manoj Srivastava to be subservient to
except that of making the method more monotonic.
I allege that Mr Manoj Srivastava activities in this mailing list
have thoroughly brought the Debain project into a serious
and unacceptable non-compliance with the social contract principle
of never concealing bugs.
Outside of Debian, the legalistic formulation of a polytope
sequence is a much relied up method for hiding the fact that the
1 winner method is not monotonic. It can take perhaps 10 to 30
years to figure that out, but less than 8 seconds if the method
is written as a polytope.
The papers: (A) (AB) (AC) (B) (BA) (BC) (C) (CA) (CB)
(A wins) = (b<c)(b<a)(...<...) or (c<a)(c<b)(...<...)
That is the Alternative Vote method's A-wins region.
It can't be monotonic since:
(a) that is simplified, and
(b) "(b<c)" does not have an A term on the RHS. Transferring
votes from (A) to (C) make A win.
I ask people to note carefully the "..." ellipsis in my (A wins)
equation. Debian has been captured by some purpose that has
nothing to do with getting the winners satisfied.
I can suspect a figurative gravely bad STINK arising from the
planning of Mr Manoj Srivastava.
What I look forward to seeing is a prompt shutdown of Mr Srivastava's
activities until this is normalised. AS far as we can tell, the
facts are these:
* there will NO checking that the method is getting better. That
is contrary to the social contract in that "bug" of not
knowing how to measure rightness, is being concealed. I even have
software only that is too hard to use that simply allows a
measurement of the nonmonotonicity of the methods. A competent
designer might be able to get other problems removed.
* there are NO good principles to Condorcet. NO PRINCIPLES that
man likes as far as I can tell. The like Condorcet but were not
led away from it by a liking of principle. A respect for principle
offers the reward of beating the others in argument, but Condorcet
likers hop away (frogs off to the ancient regime)
* there is no regard for the affect parties and there are hundreds
of individuals who are affected
* covertness and concealment of bugs has been maximised. That is
potentially able to cover a fairly evil intent. In the case of
reforming US cities. it can mean that the wrong method appears
and it takes maybe 15 years to get it removed again. The
Debian Social Contract is adverse to the concealment of bugs.
* The is NO obvious indication that well meaning members of Debian
should prefer the first of these two options:
(a) assist Mr Manoj Srivastava
(b) block Mr Manoj Srivastava, e.g. since absolutely no good
reasoning that pays close attention to every single symbol in the
old and new algorithm, is occurring.
* There is NO basis for suspecting that Mr Srivastava is involved
in a highly technical consideration in which top leaders and
others are out of their depth.
All I can think of in the way of social implications is:
* maths student Mr Voss created a website packed with errors.
Hey, I don't do that. However I never once held a purpose of
harming G. Branden Robinson or held Mr Voss' which is a purpose
of helping Mr Manoj Srivastava with his incorrect wrong changes.
>Perhaps I lost the election because too many people could not find the
>name "Brandon" on the ballot. After all, no such person was a
>candidate.
>
>Translation: PLEASE SPELL MY NAME CORRECTLY. Thank you.
There is only one winner in the game and that is the method designer.
The more precise the method then the smaller the opposition from the
super-intelligent and then usage of it can increase. No other part
of the world in the next 20 years would it.
IRV/Condorcet is a very bad single method for Debian internally when it
under an external infiltration attack on its identity from IRV wheelbarrow
pushers. Of course there are no counter arguments to what I write at
this mailing list. That is because the others are stupid and I am not.
The top leader should e-mail whomever and find out Debian is really
using the anti-incumbent IRV method. I guess the current leader can't
think about anti-incumbent type unfairness at such a great distance from
the next election, though it must be a terrible strain overseeing so
many difficult aspects in Debian running as intelligently as it is.
If there are inflitrators (evil IRV lobbyists [these ones are too
dumb to know that Condorcet is IRV hah ha ha]) then there should be
a rounding up of the leaders. No need to question them.
>--
>G. Branden Robinson | Imagination was given man to
>Debian GNU/Linux | compensate for what he is not, and
>branden@debian.org | a sense of humor to console him for
>http://people.debian.org/~branden/ | what he is.
Persons suspecting that I have quit could e-mail me at my
single-transferable-vote amailing list and maybe engage in an argument
there.
In truth there is no room for dissent on what the best 3 winner method
is (with STV style papers). I wonder if Manoj would actually provide an
argument showing that his inconsistent view of improving the system is
true).
Craig Carey <research@ijs.co.nz>
Ada 95 mailing lists: http://www.ijs.co.nz/ada_95.htm
Reply to: