[EM] Discussion n.2

Elisabeth Varin/Stephane Rouillon stephane.rouillon at sympatico.ca
Sat Mar 23 16:41:26 PST 2002

I should contact some of you soon...

Interlocutor n.2 wrote :

> Dear M. Rouillon:
> I can't follow your summary below because I can't tell what
>the points
> are trying to answer.  Also,
> you keep using the French word "circonscriptions".  That would turn
>into English
> "circumscriptions", but I have no idea what it means in discussions
> matters.  It is not a term usually used.  I think perhaps your
>terminology is
> making it hard for people to understand you, and that is why you
>aren't getting
> enough replies.

>From my point of view I use district, circumscriptions and
as synonyms.

> The main questions I ask of anyone's "model" are these:

> 1.    Are you talking about reforming the House of Commons elections
>only, or
> also the Senate, the Executive, etc.?

I have some positions about the senate and the executive but I
they can be treated as independant matters (like the length of a
or a recall procedure). So I try to focus only over the election of a
chamber of
"A different electoral system I", Stephane Rouillon, point 1).

> 2.    Does your system still have "ridings" which are the particular
> responsibility of one member of Parliament (or larger "districts"
>which are the
> responsibility of a group of MPs), or does it dispense with the
> connection altogether, electing MPs only to vote on laws and
>not to
> look after the problems of constituents?  And if the latter, to whom
> constituents turn to deal with practical problems?  Would you create
>some local
> non-party "ombudsman" positions for such matters?

As I have just explained to Interlocutor n.1,
my proposal assigns usually one member parliament per riding.
However, it could happen that the electorate of a particular riding
gets represented by a good loser form another riding. It would be the
case because of the information collected from the electorate in their
particular riding. "Re: PR and single-member districts", Stephane

In addition, I think that by removing geographical we discourage
from bribing the electorate with the localisation of a development
As in "Vote for (or against) ZLEA if you want a metro station"...
As in "Vote for sovereignists if you want an hospital"...
As in "vote for your future Prime Minister if you want a museum"...
I suppose you get the picture and have several examples of your own.

> 3  If you are going to retain the geographical connections of MPs,
> envisioning single-member ridings, as now, or multi-member
>as in PR?

I just answered that. Single-member districts that can dynamically be
to the electorate results...

> 4.    If you are envisioning single-member ridings, what sort of
>ballot and
> counting device are you going to employ?  AV by simple elimination?
>AV by
> Condorcet?  AV by some other method?  FPTP?
> (Don't go into the math details, just give me the basic name of the

I use an enhanced preferential ballot.
The voter has to rank all candidates he likes.
He or she truncates all candidates he dislikes.
So I get some information like in an approval method an some like in a
>From my point of view, there is a different position between:
A1 B2 C3 D4 and A1 B2 C3 D"blank".
D is the last preference in one case, and disliked in the other.
These cases should be treated differently and possibly generate

I do not need to rank disliked candidate because, if an election is to
a "representation exercise" nobody should be able to deny someone else
representation using any kind of negative vote. Disliked candidates
receive simply
representation. As a friend suggested to me, I added a "None" row to
list to validate a full refusal. So when I will be 111 years old,
will know
I forgot to fill my ballot. It will be treated as a Blank (white)
and not
like if I
would have disliked all candidates. Other electoral system designers
at least consider such a ballot.

For the counting method, I use AV elimination for the moment.
I keep a score for every candidate equal to number of voters
that refused, at the round he/she is eliminated,
to rally to the remaining active candidates. I do not declare a winner
at this step, just determine scores.

I have some reproaches to Alternative Vote (or IRV). It does not
preserve monotonicity and does not respect the reversal criteria.
I invite all of you to check this under
"Election Methods Resource", Demorep1.

Other methods able to find the "Condorcet Winner" when there is one
seems to
make more sense, except that none seems perfect. But I have to get
too not simply a winner. I am studying right now if Ranked Pair could

generalized in such a manner. Anyone wants to help?
I'll contact Mr. Blake Creteney and Sam Cohen (if I remember well)
for details about Weighted Ranked Pairs...
One advantage of getting scores and not a winner, I am not bothered

tie-breakers. For the moment I keep AV, it is easier to explain.

> 5.    If you are envisioning multi-member districts, what sort of
>ballot are you
> going to employ?  A vote for Party only?  (Closed list MMP) A vote
> plus for ranking local representatives?  (Open list MMP)  Can you
>for local
> MPs of a different Party, as in STV, or only within Party, as in
>And into
> roughly how many districts would you divide Canada, and how many
>would be
> in each district?

I considered voting for party, and ranking the candidates of a
district apart (Open list MMP). After a long debate, I dismissed it.
Yes, I agree that often a voter (myself) wants to vote for this party,
but thinks this candidate from another party is the best.
But I have to consider the following: what if the party defends a
that none of the candidate is going to apply (excuse me for an example
not impartial but so credible: what if all candidates of a party said
they would
abolish GST and we know none would. Would it make sense if the Liberal
(or any other) got 40% of the vote globally, but each candidate 10%
A party philosophy itself is not an option because it depends who is
going to apply it. So this is why I decided a party result should be
the average of its candidate results.

It is sad. And I would gladly consider an Open list MMP with my system
another serious proposal (feel free to defend it if you think so). But
nothing will ever guarantee me that all politicians are honest as Miss
Instead I have to rely on a district allocation without any kind of
(even geographic, or by age) to statiscally obtain equivalent samples
of the electorate, so the average of every district battle will

represent the support for a given party.

> 6.    Is your goal to ensure that the overall Party standings in
>Parliament are
> very close to the overall popular vote for each Party across the
>country, and
> does your system guarantee this goal?  Does it compromise on this
>principle of
> proportionality by any means, for example, by insisting on a
> % of the vote to keep out "fringe" parties?  If so, what minimum %
>you want
> to see imposed?

Yes it is one of my goal and it is guaranteed by the seat
There is no quota, no minimum threshold required.
Such elements are included inside systems to provide
less minoritarian goverments and shaky coalitions.

I resolve this issue differently.
I get first an optimal Proportional Representation (PR).
I am more likely to get minor governments, but again, this
is a situation that can happen with ANY electoral system
when the multiple party split the vote equally
(A tendency that seems to increase in the future because
of telecommunication and districts are less different than in the past
but this is another debate).
>From what I have seen about Italy and Israel and elsewhere,
it is not coalition government the problem (according to Mr. Trudeau
but shaky coalitions when there is 3 parties or more involved.

To ensure stable governments I suggest an option ensuring that there
will be at
one stable coalition available.
If the leading party as no need for this option, it can
process as usual. But if needed it can ask for the "crutch" option.
The way to do this is to give
the leading party a strict majority (50%+1). Some may prefer (50% or
to make sure it would be a coalition governement.
In this case, we need to add some elected to the leading party.
The ranked best losers should provide the additional elected for the
leading party.
We add the minimum to get a majority.
But it would be unfair to the other parties.
To compensate them, I suggest reducing the length of the mandate
by a corresponding fraction. If the leading party needs to double
its representative to obtain a stable goverment, make it so, but
the mandate length is halved.

> I think all these simple questions have to be answered first, before
>you get
> into the details of the math and all the technical terms like
> "circonscriptions".  I am not qualified to review or criticize the
>math, anyway.
> The others on the list are more technically-minded.  I can, however,
>give you a
> pretty clear idea whether or not I would support your system, if you
>give me
> plain English answers to the above simple questions.
> (After you have answered each question in sequence, if you wish, you
>can add a
> general summary account of the goals and features of your system, to
>put your
> answers in a context.)
> Interlocutor n.2 signature

I am quite new to your discussion list,
and I understand I sound like Mister I-know-everything.
Please, excuse me for renaming already existing methods.
I have read everything I have been suggested,
and I know my proposal can be improved.
I even know it is not the best, but it is the best
I was able to do alone.
This is why I came to receive constructive critics
and I hope some help.
Some of you have already given that to me,
I thank them.
For further goals and context, I need to translate my proposal.

> ----- Original Message -----
> From: "Elisabeth Varin/Stephane Rouillon"
><stephane.rouillon at sympatico.ca>
> Subject: Re: AV is not "the enemy"
> > Did you at least read my synopsis?
> >
> > All the AV and PR advantages you mention are available
> > within my model...
> >
> > Forget the removal of circonscriptions principle,
> > and try to understand what remains:
> >
> > 1) a preferential ballot;
> > 2) a leadership contest between candidates of the same seat,
> >  keeping their remaining votes as the electorate refuses
> >  to rally to others candidates as final score
> > 3) a fully proportional representation based on the average
> >  result of each party
> > 4) a seat allocation that maximizes the votes received by
> >   elected officials within restriction n.3
> > 5) a "crutch" option to avoid shaky coalitions.
> >
> > 6) optional: removal of circonscriptions to ensure equity
> >
> > Please.
> >

Thank you for your time,
Stephane Rouillon.

More information about the Election-Methods mailing list