# [EM] Craig Carey and None of the Rest

Craig Carey research at ijs.co.nz
Sat Feb 26 18:25:03 PST 2000

```In my opinion, this message has nothing of lasting importance in it.

At 03:51 27.02.00 , Donald E Davison wrote:
>Greetings list,
>

Mr Davison's methods of altering the number of winners maybe allows voters
to sacrifice their own interests by reducing the number of their
representatives in a chamber that is probably indifferent to that community.

STV can be modified to allow a voter to specify a vote for a special "None
of the Rest" (NOTR) pseudo-candidate, with the method allowing the counts
in the last stage that are for the NOTR pseudo-candidate, to reduce the
total number of winners by even more than 1.

For example, if the NOTR pseudo-candidate gets 3.8 times the quota at the
last stage, then the number of winners could be 3 less than the maximum
possible.

Although this NOTR preference may be able to win at an earlier stage, when
it doesn't, then that fact keeps the divisor (= denominator) of the
transfer values larger. So carefull checking of the entire algorithm and
the transfer value may be needed.

------------

>...  "If the voters are going to be allowed to reduce the
>number of winners, then why can they only reduce the number of winners by
>one?"
>
>Dear Craig,
>     The voters are able to reduce the seats by more than one in the
>methods in which they have more than one vote, methods like
>Plurality-At-Large, Cumulative Voting, and Limited Voting.

I just don't recognize those methods.

>
>Donald: >the voter can select candidates with most of his votes, but he can
>also indicate with some votes that he feels the number of seats should be
>less for this election.
>

You mean: with some preferences?. You are not writing about STV(?).
In the following I comment on where there is a single vote. Mr Davison

----------------------------------------------
Just to get this defined.
I my last message I nearly described a way to get STV to elect less than
the required number of winners. Here it is finished:

(1) Allow voters to mark a checkbox which permits them to negate their vote.
(2) During the counting, negative counts are removed by adding the same
count fully equally over all the permutations needed to get that count
non-negative.
(3) The election is solved by whatever preferential voting method is used
(4) The method needs to provide a count for each winner after all the losers
subtracted and then all candidates with a negative quantity of votes
are regarded as being losers.
----------------------------------------------
Possibly people who cast negative votes would hope that the number of winners
was not reduced.

...
>     Besides, we cannot ask the question because we do not know the full
>question - we do not know how much of a reduction to ask for in the
>question.

(Is this being excluded?: the voting papers can't allow selection of 3
candidates ("please tick the checkboxes and select the 3 candidate you
want") only because the method might later elect only 2 candidates.

That would be just a case of voters not getting what they wanted at an
election.)

[1]  (A  B  C  D  NOTR.  _  _  _)
Why not allow to voters to cast a preferential vote and allow them their
maximum power and have the "None of the Rest" option be a pseudo
candidate. Good methods may not be defined when NOTR is not like an
ordinary candidate.

[2]  (A  B  C  D  NOTR.  _  _  _)
Alternatively, a "None of the Rest preference" (NOTR) preference would
have the same power for each voter irrespective of whether their main
vote for candidates helped elect candidates or not.

...
>Closed Party List:
>    In this method the voter's lack of choice cannot be corrected, except
>by using Open Party List, but in this method the voter would first vote to
>determine if or not the seats for each party cannot be more than one half
>the number of candidates a party has running, for this one election only.

Reducing the length of the party list can damage proportionality, unless
the insufficient quantity of MPs taken out of the party list is "topped up"
with extra MPs from electorates, or....

>
>Open Party List:
>    This method will give the voter far more choice than Closed Party List.
>    The voter would first vote to determine if or not the seats for each
>party cannot be more than one half the number of candidates a party has
>running, for this one election only.
>    How slots for None of the Above are to be used will depend on whatever
>method is used by the voters to determine the order of the party lists.

>    The seats that are not being filled will be a problem to the parties,
>but the answer to this problem is for the parties not to control which of
>its members can run for office. Parties should be putting up twice as many
>candidates as they expect will win. Any dues paying, card carrying member
>of a party should be free to run for any office and should receive equal
>support from the party. It should be up to the party voters to decide which
>candidate is to be elected.

Parties may prefer to run numerical simulations rather than using your
figure of two (or anybody's figure of 'two').

>    In the case of small elections, it should not be any more dificult to
>become a candidate and run for office than it is to apply for a position of
>employment.
>

Vote wasting by the method makes it reasonable for parties to limit their
number of candidates. But Mr Davison was perhaps referring to independent
candidates.

In the earlier Droop war, a method with a large transfer value, that is
larger than what is obtainable, did cast some implied shadow of corruptness
over the selectors who perhaps didnt learn at that could have been learnt
from the reformer. I.e. a Hare quota for winnes is better than a Droop one.

But in the paragraphs just above, this is written: "Any dues paying, card
carrying member of a party should be free to run for any office". Surely
that is a basis for an opinion that Mr, Davison would prefer transfer
values to be big (i.e. the method to be good).

An example of this:
if an Irish party is fortunate enough to have loyal supporters who won't
vote for a candidate that is not of that party, then that party could get
less seats if it put up 50 candidates, than what it would get if it had of
put up 8 candidates. STV would create such a problem and an optimal method
may too (since the problem appears in 3 candidate elections if I recall
correctly)

[Another fancy idea is to allow voters to determine the number of months
that a candidate (e.g., a senator, or member of congress) may remain a
representative.]

```