[EM] Re: IRV vs Plurality

Chris Benham chrisbenham at bigpond.com
Tue Aug 5 17:49:02 PDT 2003


  James G-A  and others,
I  applaud the recent posts in this thread from  James Green-Armytage 
and John Hodges. I am all for Proportional Representation, and I agree 
that it is completely ridiculous for proponents of  Approval, IRV and 
Condorcet  to be in any way publicly getting in the way of each other's 
reform efforts (in effect helping Plurality to survive). I have some 
definite opinions on the relative merits of several single-winner 
election methods  and I shall now lay out my case.
For a start I think all  good  election methods  should allow and be 
able to handle  the voter fully ranking all the candidates. On being 
offered a voter's full ranking a method should either respond :
 (a) "Fine, I can handle that, that's all I need"  (ranked ballot 
 methods like IRV and  Condorcet, and also strictly speaking Plurality).
or
 (b) " I will accept that and probably that  is all I will be interested 
in, but if you care to provide some extra information on how you rate 
the candidates  then  I may have a use for it"  (ranked ballot with 
approval cutoff, or high resolution ratings ballot)
but  NOT
 (c) " I can't handle that and I won't accept it. I  will only accept 
some arbitary number of rankings/grades which is less than the number of 
candidates, so you deal with that and then I am going to pretend  that 
 this arbitary number represents your true number of  preference levels" 
 (take a bow, Approval and MCA).

My three most fundamental standards:
1. Condorcet Loser:
 With sincere voting  it must not be possible for a candidate to win 
which, in pairwise comparisons, every other candidate is prefered by a 
majority. Completely unacceptable scenario: A  despot  supported by only 
a small proportion of  the population  is finally forced to submit to a 
"free and fair"  election. Lots of candidates  stand, everybody votes 
for their favourite, and  the despot wins.
IRV  passes;  Plurality fails (spectacularly), and so does  Approval 
 (strictly speaking).

2. Majority Favourite:
With sincere voting, a single candidate who is the favourite of the 
majority, must win.
IRV , Plurality and MCA pass. In my book, Approval doesn't.

3.Mutual Majority
With sincere voting, if  a faction of voters comprisng  a majority 
prefer a set of candidates over all other candidates, then one of the 
candidates in the prefered set must win.
IRV  passes;   Plurality, Approval and MCA all fail.

Another important one is:

4.Monotonicity:
Changing nothing except ranking a candidate higher on some ballots must 
never change that candidate from a winner to a loser, and changing 
nothing except ranking a candidate lower on some ballots must must never 
change that candidate from a loser to a winner.
IRV is "famous" for failing;    Plurality, Approval  and nearly 
everything else passes.

I am middle-aged and  I have lived in Australia all my life, in one of 
 the majority of  States which elects the  "lower" house (on which the 
government is based) by IRV  in single-member disricts. As far as I can 
tell, noone even notices that IRV is not monotonic, nor do they notice 
the  small  Lesser of  2 Evils problem.
I might now digress to point out some of the biggest differences between 
the political cultures of  Australia and the US.  One of the two major 
parties here is called the  Australian Labor Party. It  is analogous to 
the Labour parties in the Britain and New Zealand.
It  was originally  started  by the trade unions (labour unions), and it 
still has links with the trade union bureaucracy. By contrast with the 
US, Australian politics has been polarized  largely along class lines as 
pro and anti  Labor. There is more party loyalty than in the US.
Another  huge difference is that in Australia voting is compulsory. 
There is very little opposition to compulsory voting, and nearly all of 
that is seen as coming from self-interested  anti-Labor parties. Also in 
most Australian IRV elections, voters are legally obliged to put a 
number in every box.
Voters all either vote sincerely, or they sincerely choose one of the 
political parties which is contesting their district and then just copy 
that party's "how-to-vote card". Strategic or  "tactical" voting is 
almost unheard of  and doesn't interest anyone. There is a small centre 
party which in a number of seats would be the  CW, and one of the 2 
major parties in some of those seats might be (based on polls and the 
results of  the previous election) seen as having no hope of winning. 
The supporters of that party never think "Our favourite party can't win 
this seat, so lets abandon/"betray" our favorite and try to elect the 
lesser evil". I would say that those voters who are interested in 
elections would nearly all regard it as part of their civic duty to vote 
sincerely, and those that aren't interested, aren't interested. Voters 
are not just concerned about which or how many seats are won by which 
parties, but irrespective of that they are also interested to see their 
favourite party get as big a share of the total vote as possible. In the 
State of  South Australia, where I live, there is an institutional 
 attempted anti-gerrymander. There used to be a deliberate policy of 
having  fewer voters in  the relatively sparsely populated rural seats, 
but now  that is seen as completely unacceptable and offending the 
fundamental principle of  "one-vote-one-value".
The Lesser of  Two Evils problem.
The vote splitting problem under Plurality  has a profound effect  on 
 the nominations. As soon as there are more than two candidates, the L2E 
problem becomes painfully obvious, and the door opens for the Condorcet 
loser. The individual voter who abandons Lesser Evil frontrunner for 
 no-chance Sincere Favourite, is faced with the horror of   helping 
 Unacceptable frontrunner. So the elections become like self-fulfilling 
prophecies: the 2 major parties are the 2 major parties partly because 
everyone thinks they are.
In huge contrast to this, L2E only becomes a concern under IRV  if 
 there is a possibility of both of  two things happening:
1. Your favourite overtakes Compromise, and then
2. your favourite loses to Unacceptable , when Compromise would have won.
This is never a problem in a 3-candidates election in the situation 
where your Favourite and Compromise are close allies, and tightly 
exchange preferences. This happens in Australia in some seats: there are 
two anti-Labor conservative parties who presently are in federal 
government as a coalition.
So in my book those despised US  IRV advocates who "falsely" claim that 
IRV completely gets rid of the L2E problem are only guilty of slight 
exaggeration. Moreover it seems to me that CVD is proceeding according 
to a reasonably intelligent 2-stage plan.
By far the best  and fairest way to open up electoral politics for more 
parties is to use PR. Of all the PR systems currently in use around the 
world, the one which gives the greatest power and flexibility to the 
voters is  ranked-ballot  STV-PR. The idea is to first introduce the far 
less radical  IRV, which will result in (a) voters becoming accustomed 
to a ranked ballot , and (b) other parties nominating, at least becoming 
visible and then to start winning significant (say 10-25%) numbers of 
votes. Then the stage will be set to push for STV-PR , and maybe 
Condorcet for single-winner elections. If  Approval is introduced, then 
there is a danger that the voters will get too comfortable with 
single-member districts, and PR will never be achieved.
So to sum up, Plurality is a completely unacceptable nightmare. In 
comparison IRV is not a nightmare and is infinitely better, although 
certainly flawed and not as good as Condorcet. Changing from  Plurality 
to IRV  is  therfore a very worthwhile reform!

Chris Benham




















 







3.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20030805/20cd59cf/attachment-0002.htm>


More information about the Election-Methods mailing list