NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/can/politics/9819

Internet Message Format | 1992-11-18 | 5.5 KB

Path: sparky!uunet!uunet.ca!synapse!lucien.saumur
From: [e-mail] (Lucien Saumur)
Newsgroups: can.politics
Subject: Re: Proportional representation
Message-ID: <[e-mail]>
Date: 18 Nov 92 20:24:00 GMT
Distribution: world
Organization: SYNAPSE BBS - GATINEAU, QUEBEC - 819-561-4321
Reply-To: [e-mail] (Lucien Saumur)
Lines: 125

From: [e-mail]
To: can.general,can.politics,soc.culture.canada

In article <[e-mail]> (Ross Wetmore) writes:
>In article <[e-mail]> (Gerry Gryschuk)
writes:
>>From article <[e-mail]>, by
[e-mail] (Mathew
B
Englander):
>>> [e-mail] (David Adrien Tanguay) writes:
>>>>What are the cons of a transferable vote system? (i.e., where the
voter
ranks
>>>>all candidates, somewhat similar to the way the party leaders are
chosen)
>>>>It would seem to preserve the ideological direction of the
electorate, yet
>>>>limit the representation to the more popular parties, helping to
avoid
>>>>parliamentary deadlock.
>>> The greatest problem is that candidates can be hurt by an
*increase* in
>>> support. For example, suppose that candidates A, B, and C are
running
>>> for the party leadership. Suppose that there are 675 voting
delegates
>>> divide roughly into four groups:
>>>     Group I: 150 voters, whose ordered preference is A, B, C
>>>     Group II: 100 voters,  "      "        "      "  A, C, B
>>>     Group III: 200 voters, "      "        "      "  B, C, A
>>>     Group IV: 225 voters,  "      "        "      "  C, A, B.
>>> (What I mean here is that, for example, Group II prefers A to C,
but
>>> prefers C to B -- and thus presumably prefers A to B.)
>>> Now, what happens in the vote? On the first ballot, the results
are
>>>     A: 250 (Groups I and II)
>>>     B: 200 (Group III)
>>>     C: 225 (Group IV).
>>> No candidate has a majority, and since B has the lowest total he
is
>>> eliminated and A and C have a "runoff". Note now that the 200
voters
>>> in Group III will now vote for their 2nd-favourite candidate, C.
>>> So the second ballot results are
>>>     A: 250 (Groups I and II)
>>>     C: 425 (Groups III and IV).
>>> And C wins the leadership.
>>> So far, so good. But what would have happened if, before the
first
>>> ballot, C did some last-minute campaigning and won over all the
members
>>> of Group II (which gets renamed Group II*). The breakdown of
groups now
>>> becomes:
>>>     Group I: 150 voters, whose ordered preference is A, B, C
>>>     Group II*: 100 voters, "      "        "      "  C, A, B
>>>     Group III: 200 voters, "      "        "      "  B, C, A
>>>     Group IV: 225 voters,  "      "        "      "  C, A, B.
>>> Since C has increased her support, and she would have won even
without
>>> this increase, we would expect her still to win now. But does
she?
>>> The new first ballot results are:
>>>     A: 150 (Group I)
>>>     B: 200 (Group III)
>>>     C: 325 (Groups II* and IV).
>>> So A is eliminated, and Group I supports its second choice, B, on
the
>>> second ballot:
>>>     B: 350 (Groups I and III)
>>>     C: 325 (Groups II* and IV).
>>> So B wins! The shift of voters *toward* C actually caused C to
lose
>>> the race. This is why this sort of voting is fundamentally
flawed.
>>
>>Sorry I don't normally leave in the whole post just to respond a
little at
>>the end but I think it's necessary here.
>>Anyway, yes this method would be flawed. However a simple fix would
be to
>>set a fixed number of "points" to the place a person voted a
candidate.
>>Simply base it on the number of candidates running and no problem
should
>>arise. That is in your examples a first place vote is worth 3,
second 2 and
>>3rd 1. In this case C wins both times and by a wider margin in the
second
>>scenario as she should. The math is left as an excercise for the
reader.
>
>  I should have gone back to the original post by Matthew, but it
has
>flushed :-)
>  Since in the example, as has been pointed out, any pairwise runoff
>would yield a different result, there is no strong preference in the
>electorate and choosing any single candidate should perhaps best be
>done by a lottery technique in such pathological cases :-).

     I do not think that this is a "pathological case" but a
legitimate case where the voters have no strong preference and where
the winner should be rightly decided by a lottery technique. But I
propose that such cases would be rather rare and of little importance
for our purpose.
     What is of the ultimate importance, in a democratic society, is
that the electoral system serve to elect the candidate who is
preferred by the majority of the voters, when such a preference
definitely exists, as is usually the case. To do so, it is important
that
     * the voting system allow the voters to rank the candidates and
       that
     * the vote compilation system does not eliminate any candidate
       on the basis of the first preferences of the voters.
     This is so because the preferred candidate may be no one's first
choice but everyone's second choice so that he may in fact be
preferred to every other candidate and deserves to be declared
elected rather than eliminated.
     The vote compilation system should consider the election as a
number of two-candidate sub-elections where each candidate is opposed
to every other candidate and where the candidate who wins in every
sub-election in which he is a party is declared elected.