[EM] Craig: I did include ballot-counting code

Craig Carey research at ijs.co.nz
Tue Feb 1 05:08:58 PST 2005



MIKE has got me black-listed so, as with Mr Piaelli, I can only reply publicly
and I can never cut out the step when MIKE makes messages be diluted and
verbose by expressing false and untrue ideas.

In this matter you were simply in the wrong: it is a dull brainless piece
on how you hold a great desire to be sensitive to the public's desire to
"express rankings". What happened between 1992 (you were a CVD member) and
1997 (when believer in pairwise comparing) ?. At some point you got
attached to lies on what it was that voters want. They fill in ballot
papers with preference lists and hope for a correct set of winners.

Your programming idea, described with the words "express[ed] rankings"
is totally different from the voters' idea of:
  (a) ballot paper counts as input, and
  (b) sets of winners as output.

In 2000 you were a person who would never ever say (privately AND publicly)
that there was only "one" input to the method function returning a winner
set: the "collection of ballot papers".

After 4 years of thinking about whether voters in an IRV election only
create a collection of IRV ballot papers, what you have got for us is the
total lie, "no", since they instead produced "express[ed] rankings".

Below you say that Russ probably argued that Americans don't actually
draw antisymmetric matrices onto their ballot papers. We knew that.

Making false claims that a public wants to have a collective preference
is done to protect one of your beliefs. A public does not "express rankings"
but instead the falsity of the claim is a stupid strict designed to
conceal a wrong belief you have.

The public fills in preference lists. What your Python code did was designed
to do was to trash the public into doing something it never did, sought, or
has a reason for: create a pretext and deceptive cover story for you. What
do you want: to believe the lie which is that Condorcet's original thought
was correct and that there seem to be Comdorcet cycles (those difficult
to resolve things).

The reason the input stage code was missing, is to help you get attached.
E.g. some concrete to keep you at the bottom of the water off the edge of
the wharf, and eventually you would be cataloguing the fish and boots.



At 2005-02-01 04:03 +0000 Tuesday, MIKE OSSIPOFF wrote:
>
>Craig Carey said:
>
>What the blazes ?: OSSIPOFF's Python code couldn't even accept ballot paper
>counts as input. That is what I saw at the Piaelli website.
>
>I reply:
>
>  Very good point. I agree that it didn't make sense for the Python listing 
>to not have code for receiving and counting rankings. The Python code that 
>I'd written, and Russ had it in its final debugged version, received and 
>counted the raw input to get the pairwise vote totals.
>
>  I wanted the program to be usable by anyone who wanted to copy it, and I 
>wanted it to be complete and self-contained. So I included code to receive 
>the rankings from a keyboard, and, from those rankings, to determine the 
>pairwise vote totals, and use those to determine the winner.
>
>  Russ passionately insisted no one would want to enter rankings from the 
>kekyboard. One could ask, then, if he thought that people would rather 
>determine the pairwise vote totals without any assistance from a computer 
>program.
>
>Sure, for a user who writes programs, that user can write his own 
>ranking-receiving and counting program. But I wanted the program to be 
>accessible to people who weren't inclined to write their own input code. No 
>such luck. Russ left out the keyboard input and count code, leaving a 
>program that had the pairwise vote totals as its inputs--inputs to be 
>obtained by the user however s/he manages to.
>
>Of course someone who wants to write their own input code could disregard 
>the keyboard-input code.
>
>  By the way, the code to receive the rankings from a keyboard and count 
>them to find the pairwise vote totals was considerably longer than the part 
>that uses those pairwise vote totals for the BeatpathWinner algorithm. 
>That's because, for practical use, it's necessary to give the user a way to 
>correct any keyboard-input errors that s/he makes, and to indicate when s/he 
>has completed each ranking, and when s/he has entered all the rankings. That 
>makes the code considerably longer.
>



That is so seemingly dumb, but it is really there to help OSSIPOFF prop up
some belief. Also the keyboard errors argument seems to be a lie. Though
OSSIPOFF can't stop a wide range of blunders, it is to be expected that the
public can't start making errors.

The best solution is for Piaelli to close MIKE's webpage.

We are talking about the format of the input data. I have considered the
problem and I conclude that computer programmers who develop algorithms,
will prefer this input:

 Table A
 ----------------------------------
   (1, (1, 0, 0, 0))  % = 1*(A)
   (2, (1, 2, 0, 0))  % = 2*(AB)
   (3, (2, 1, 0, 0))  % = 3*(BA)
   (-4, (3, 2, 1, 0)) % = -4*(CBA)
 ----------------------------------

In instead the algorithm gets IRV ballot counts in a tree stuctured form,
then the algorithm can be much harder to understand.

OSSIPOFF;s 3rd option, of telling users to sum some of their STV papers,
is the same as requiring that input data be multiplied by a matrix.

For the input to be standardized, then every one else's algorithm has
to start off by multiplying by the inverse of MIKE OSSIPOFF's matrix.

The motive for creating the matrix seems to be so muddy and dark, that MIKE
is only prepared to withhold the motive and mislead about the motive, when
saying that the public will make mistakes.

Incredibly, MIKE has no words defining the matrix, and no words saying if it
can be inverted.

He has only got ONE program: surely it might as well set a good example,
for other programmers to follow.

Voters don't create cycles with their voter. They don't have preferences
to express, i.e. clues on what the method should do when their vote is
mixed with other votes. MIKE made a false claim saying voters expressed
preferences.

The title of this message

    "Re: [EM] Craig: I did include ballot-counting code"

is totally untrue, as I recall, since MIKE never had code to count the
first column of numbers that is in Table A.





More information about the Election-Methods mailing list