I told Rob Richie, use "IRV", its catchy (was Re: [EM] Approval Elections & Effective Weights

[Craig Carey]

At 2001-04-11 15:15 +0100 Wednesday, Martin Harper wrote:


==================


 >
 >To: election-methods-list at eskimo.com
 >Subject: [EM] Approval Elections & Effective Weights
 >
 >Replies mainly in the text.
 >
...

 >Incidentally, you complain about me using the term "IRV" - you
 >say: the so called 'IRV' method
 >
 >I thought that this was the standard name: "Instant Runoff
 >Voting", or "IRV". What do you call it?
 >

The term IRV is a minor term that is hardly used except in parts
of the United States of America. I am not aware that it is used
in the mathematical literature. The CVD does not like mathematics.

It is just not a term that is respectable mathematically.

When I once browsed to the CVD website and read about what
Rob Richie said to either Senators or Congressmen, it didn't
get mentioned that the CVD made up the dialect (i.e. not part of
BBC English's vocabulalary), "IRV" / "Instant Runoff Vote".

Mathematically it is fast sinking ship (at my PaP list is an
example showing that it will give a power of 2 to a single IRV
vote -- it is hard to guess what the fact is, but it seems the
truth is that the power flops out rather than being perfectly
constrained and I suppose it is perfectly fair to say that STV
is going to get badly sucked down and have its credibility
destroyed primarily because it has the CVD IRV method embedded
in it -- all sorts of faults can be tolerated but not that
one, but we have not heard the politicians say that...) That is
just my opinion of the moment.

Here is a private message I once got [with deletions done
by me]:

At 2000-05-04 22:08 -0700 Thursday, Gary D. Shapiro wrote:
 >...
 >> Santa Barbara.
 >>
 >...
 >>So in the meantime, groups such as ours put Choice Voting and
 >>IRV on the table. After all, I'm the guy who suggested the
 >>term "Instant Runoff Voting" to Robert Richie (on 2/29/96).
 >...
 >
 >>Gary Shapiro                  <http://www.garyes.com/email/>

The month and the day of the day of the week are swapped and
disordered so it certainly seems the message is as authentic as
I know it is.


 >Sadly, my pet psychic was on holiday the day I answered Mr.
 >Davison's question, so I had no option but to answer the question
 >he actually wrote, rather than what might have been going on in
 >his brain.
 >

Unlike other subscribers, there is less ground to traverse to fix
up the errors in Mr Davison's texts, I have so far noted.

...
 >I believe I specifically said that it was a uniform probability
 >distribution. IE, there was a 50% chance of there being one 'B'
 >voter, and a 50% chance of there being no 'B' voters.
 >

Where are the formulae and all the steps of the mathematical
argument making the final case for that conclusion?.

 >> Unfortunately probabilities associated with utility theory are a well
 >> guarded secret.
 >
 >You really ought to tell Blaise Pascal that - he invented them
 >back in the Rennaisance some time. By now, they're comparatively
 >common knowledge - taught in schools and such, and used to help

Nothing can be taught here without extreme and exacting lucid
rigour. We shall assume associativity of addition and so on...

 >> The "effective weight" is defined by me so that it ends up with
 >>the meaning  that people would think it ought have: it is the
 >>ratio x/y, where x is the  number of FPTP votes that are required
 >>to offset y of another vote.
 >>

There would not be an 'effective weight' if there is not a tie
between two winner sets.

...
 >EG: As you know, Plurality (aka FPTP) has a spoiler problem, which
 >IRV largely eliminates. For example, a Nader supporter in the last
 >USA election might vote Gore in Plurality, to attempt to keep out
 >Bush. In IRV, however, they don't need to worry about that spoiler
 >problem, so they can safely vote Nader>Gore instead. On the other
 >hand, a Gore supporter would vote Gore in Plurality, and Gore in
 >IRV.
 >
 >This is going to be the case in all methods: a FPTP vote for a
 >candidate might be translated to a number of possible votes in more
 >advanced methods. How do you intend to resolve this?
 >--

I completely ignore that. Checking of the method that results can
occur. There could be backtracking of some sort but I don't see
that imposing the very good rules that the "IRV" method fails
would lead to a good enough method.

 >And finally.
 >
 >Don wanted to know if Approval was used in any real elections, and
 >I answered that question.

You do not agree that he meant or nearly meant what I said he
meant. Maybe I should have said that the elections had to be
repeating as well as governmental, large, and public, and used in
a nation. It would be no good to find the name of a nation that only
used positive-support ASVB once and never used it again.

 >                      Don also wanted to have details of actual
 >results, and I replied that I didn't have those details, so he'd
 >have to look for them himself. There is a further question now:
 >is/was Approval used in any public, country-wide, elections?
 >
 >According to the same paper I shamelessly plagiarised last time for
 >details of elections to professional bodies like the IEEE:
 >
 >> "In the former Soviet Union, many elections involved the
 >presentation of a list of candidates to the voters, and voters were
 >only allowed to cross names off the list: this system is equivalent
 >to allowing the casting of approval votes (for the candidates not
 >crossed off)...... In 1991 Oregon conducted a public referendum
 >involving five alternatives using approval voting."
 >
How can we be sure you are not describing the Cumulative Vote.
(or a variant, e.g. the newly coined "4/11/2001" ASVB method, which
I am pleased to say, has been much commented on at this list.



 >So here we have the former Soviet Union using "Disapproval", which
 >is equivalent to Approval, and Oregon using Approval for a public
 >referendum. I do not know whether there have been any public
 >country-wide Approval elections since then - I've not looked or
 >particularly cared. As in my reply to the last question, I do not
 >have the results for any of the mentioned elections: primarily
 >because I don't care. Does that answer your question, Craig?

I do not know which "Approval" variant Mr Harper is writing on.
Perhaps he would help us all out by posting up an image of the
ballot paper. Some forigners will be starting to get suspicious:
Florida had ballot papers, and that is just one state.

I presume states do not use Approval since it allows to moustached
male voters who wanted 4 candidates and voted for 3, a lot less
power than a row a kilometre long of hundreds of overweight ladies
that emerged from strengthened low buses and who agreed at their
earlier bingo/lotto meetings to vote for nearly 1/4 of the 20-50
candidates. Alternatively the story could be altered a little and
the women could be replaced with the CVD and it staff that put
pamphlets into envelopes.

Can you you tell Mike Ossipoff to make available that function
that is the exact probabilities that are used in the utility
theory calculations. It would be a function of a function, wouldn't
it Mr Harper?. I understood the explanation it got lost with the
history of ancient France. Maybe Mike has redone the mathematics
and in private e-mail he could send it to use and by that simple
strategem his steely intent to keep the list uninformed of why an
inferior version of the ASVB method, is a method that anybody in
any nation, would want to use.

I will follow the topic with interest. The British once calculated
that the universe had 10**98 atoms or something. I questioned the
popping in and out of the number 0.5, and the defence is to have
a mathematician pop in, along with a lot of schools. If it gets
worse and galaxies/stars start to pop in and out, the readers of
the list would begin to tiresomely suspect you really were
learning at the feet of the Approval supremos.
The centre of the universe certainly could be tentatively
suggested to be the boulder.bcn colorado website. Given how little
it changes, I have not really looked at since 1996.


> > Unfortunately probabilities associated with utility theory are
> >  well guarded secret.
>
>You really ought to tell Blaise Pascal that - he invented them
>back in the Rennaisance some time. By now, they're comparatively
>common knowledge - taught in schools

easy on, <<"U. S. Schools">> - in other lands, maths is actually
taught in maths classes and their public believes that. Martin
continues...

>and such, and used to help
>make business decisions and suchlike under the pseudonym of
>"Expected Value". --

In other words, the function must exist and the argument was that
a function with an integral of 1 can be multiplied by x and
integrated over x. Can you flesh out some of the missing parts
of your defence of the 0.5 value that appeared in your exposition
to me, of utility theory or of, of Mike's defective since ill-defined
variant of my ASVB method. I invited private comments on how to get
this ASVB method standardised.