[EM] Discussion n.1

Elisabeth Varin/Stephane Rouillon stephane.rouillon at sympatico.ca
Sat Mar 23 16:39:44 PST 2002

This seems not far your current discussions...

>No M. Rouillon, you are wrong. You fail to see that your idea is not
>PR, it
>is simply an extension of the Borda count that gives representation
>second, third, etc. place winners.

Yes, it seems we do not share the exact definition of PR.
You seem to include in PR only the traditional PR model
using the whole country like one unique district as it
is done in Netherland... Maybe you consider STV a less
PR acceptable model (but still a PR model because it uses multi-member
>From my point of view, a sytem is a PR (Proportional Representation)
system as much as it can minimize the following norm:
1/2 Sigma | Xi - Si |               (1)
Xi represents the suffrages (fraction of the vote) party i received,
and Si the number of seats party i received.
I have seen somebody else on this list use the same definition
without the 1/2 factor.

STV is broadly proportional. A Borda count with representation for the
would be too, because the relative weights according to Borda waste
the relative votes information.

My model is far from a Borda count:
1) weights are different,
2) at the seat allocation step,
    a second could be elected when a first could be eliminated.

By the way, I have searched everywhere according to
Mr. who has seen everything under the sun, and it seems
2) is ont of the properties of my proposal that no current
model has brought out of the shadows...

>How far down are you prepared to go?
What do you mean by that?
If you are asking were I put a quota for representation,
there is none, because it is useless. If an election should
be a representation exercise, any philosophy (or party) able
to gather enough vote for one seat should get it.
So my proposal still allows independant runnings
which is a good thing for denouncing dishonest politics.

>What you have in effect with several candidates being elected is a
>single-member constituency being converted into a multi-member one,
>is not PR but rather the semi-PR system of the single
>Interlocutor's signature

The scores obtained by each candidate helps generate a global
distribution of seat respecting an optimal Proportional
Norm (1) is equal to its minimal value, as with the Netherland model.
The difference is that instead of using a list ranked by a couple of
from the party ( a kind of oligarchy because first names almost all
the electorate builds the list through the election itself.
And for your last remark, my proposal usually does not lead to several
being elected in single member constituency: the average number is
always one per district.
Rarely, there might be none, two and in very exceptional case more
two elected per
I agree that my model tends to convert some single-member constituency
into multi-member
constituency: the ones needing to transfer an elected between each
other. But this is a quality.
I can give you an example. Suppose 2 districts with only two
each, and the following results:

District 1
Party A candidate   51%
Party B candidate   48%
None                       1%

District 2
Party A candidate      2%
Party B candidate      3%
None                       95%

I did put None as I described within an enhanced preferential ballot
because sometimes all a voter wants to say is that all candidates
are bad. When two good candidates rip appart each other
while other well-known incompetent (according to the electorate

just do not fit, what should we do? A clever human observer would
that if the refusal comes not from the district electorate nature (or
it comes from the candidates aptitudes... Hence, candidate of party B

or victim of the other party strategy. But as a matter of fact, the
electorate is
a victim too and not party B because it will still get one elected.

As electoral system designers, we work very hard to counter
Should we not do the same to counter strategical district allocations
In such a previous case, the good thing (from a voters satisfaction
measure point of view) is to declare both district 1 candidates
and none from district 2.
This too is getting a fair vote in Canada.

My system is able to recognize and solve such a case. Does any other?

Finally, if you know any website describing the semi-PR system of the
single non-transferable vote
(SNTV), I would be pleased to look at it. But I doubt it is the same.
As I described, my model uses a preferential ballot and allows to
transfer a vote
within the single-member constituency.

At least it seems you have tried to understand my proposal even if
I did not translate it yet.
I would be glad to learn more about your option and hope for more
constructive critics.

Stéphane Rouillon.

More information about the Election-Methods mailing list