[EM] Nightmare On IRV Street ?

[Dave Ketchum]

On Wed, 2 Jul 2003 20:19:10 EDT Dgamble997 at aol.com wrote:

> 
> Adam Tarr wrote
> 
> A>B>C 42%
> B>A>C 5%
> B>C>A 7%
> C>B>A 46%
> 
> So here we have a more realistic "turkey" scenario.  A candidate with a
> core support of only 12% manages to win an election. 


As easy to look at the other side.  What percentage of voters would be 
most DISpleased by each possible winner?
      A 53
      B ZERO
      C 47

Clearly this is a case of Condorcet minimizing UNhappiness.

Take another look at A voters' choices after doing the most they can for 
their candidate with their own votes:
      B>C - as above, they prefer B over C, and thus help B wen.
      truncated - they care not between B and C, leaving the decision to 
other voters if A does not win.  Here C wins, 46-12.
      C>B - they prefer C over B and C wins - proper vote ONLY if this is 
their true second choice.

I read, sometimes:
 >
 > Under Condorcet by casting a second preference for compromise candidate
 > B both A and C voters have effectively defeated their first choice and
 > elected B.
 >
NOT true:
      First, it does nothing to weaken their first choice.
      Second, if the C voters can con the A voters into not making B the A 
second choice, C has a better chance of winning.  True, but the A voters 
are stupid if they vote anything other than their own preference as to B vs C.

I also read of "quality".  A voter may be able to have some such thoughts, 
BUT there is no way to say anything about such in a Condorcet vote.  The 
illustrations suggest an A voter may like B trivially better than C - but 
the A voter can and should vote the same way if liking A trivially better 
than B, and each much better than C.

> Compare that, if you will, to the nightmare scenario of IRV:
> 
> 10% FarRight>Right>Centrist>Left>FarLeft
> 10% Right>FarRight>Centrist>Left>FarLeft
> 15% Right>Centrist>FarRight>Left>FarLeft
> 16% Centrist>Right>Left>FarRight>FarLeft
> 15% Centrist>Left>Right>FarLeft>FarRight
> 13% Left>Centrist>FarLeft>Right>FarRight
> 11% Left>FarLeft>Centrist>Right>FarRight
> 10% FarLeft>Left>Centrist>Right>FarRight
> 
> Centrist has the most first place votes, the most second place votes, and
> the most third place votes.  Centrist is the only candidate who does not
> appear fourth or fifth on any ballot.  Centrist would win in a landslide
> over any other candidate in a two-way race.  Centrist is quite obviously
> the popular choice by ANY reasonable measure.
> 
> In Condorcet, plurality, top two runoff, or really any reasonable method,
> Centrist wins.  But in IRV, Centrist is eliminated before the final runoff,
> and Right wins in a squeaker.
> 
> To me, the IRV nightmare scenario is obviously far worse.  In my nightmare
> scenario for Condorcet, no voter has any real reason to regret their vote
> after the election, while in the IRV vote a solid third of the electorate
> have good reason to regret their strategy.
> 
> 
> Consider the following plurality nightmare:
> 
> Florida November 2000
> 
> Bush     2,912,790
> Gore      2,912,253
> Nader    97,488
> 
> Bush wins the 25 electoral college votes of Florida and thus the presidency.
> 

Actually, the US Constitution left the Florida Legislature wide discretion 
as to how their EC delegation should get determined.  Choices include:
      Plurality, as actually done.
      Ranked voting, as some seem to promote, likely going all for Gore in 
this case.
      Let voters vote as in Plurality, but distribute EC membership 
proportionally:
           In this case 13 Bush; 12 Gore.
           One for Nader if he had a few more votes.  If there had been a 
possibility of one or two Green EC members, the Greens could have had 
them committed as to what to do if there had been no possibility of 
actually making Nsder President.


> Or the top two runoff nightmare:
> 
> First round of the French presidential election 21.04.2002
> 
> Chirac (RPR, right)  19.9%
> Le Pen (FN, extreme right)  16.9%
> Jospin (PS, left) 16.2%
> other left  ( 6 candidates) 21.3%
> other  ( 7 candidates)   25.6%
> 
> Divisions between candidates on the left propel a candidate of the 
> extreme right into the second round ensuring Chirac's victory as the 
> only acceptable candidate. A 3% fall in Chirac's support could have lead 
> to a Le Pen / Jospin run-off leaving the main stream right without a 
> second round candidate.


A SOLID indictment of this runoff approach - HARD to do runoff without 
some invitation to disaster.  Has this convinced the French they need to 
change their approach?

> 
> The Condorcet nightmare :
> 
> 470 A>>>>B>C
> 40 B>A>C
> 30 B>C>A
> 460 C>>>>B>A
> 
> B a low utility compromise candidate wins.


This was discussed above.  I claim this is NOT a nightmare.  While the A 
and C voters were extreme partisans and desired to win, neither got the 
votes.  They each voted that B was their preference for second choice over 
helping their worst enemy win.

> 
> All single seat methods are capable of producing bad results.
> 
> This is why I believe that single member methods should only be used for 
> single offices ( mayor, governor, president, etc) and that multi-member 
> bodies should be elected by proportional representation. There is 
> nothing and can be nothing that is proportional about the allocation of 
> a single seat. My preferred single member method is IRV.
> 

As should be obvious, I prefer Condorcet:
      The above IRV disaster is a true disaster, to be run from.
      If the above is the worst that can be said of Condorcet, it wins.
      I WOULD make it part of Condorcet that the voter is permitted to 
give multiple candidates (two or more) the same rank.  When counting votes 
for a pair that share a rank:
           No effect on scoring these vs other candidates.
           Count half a vote for each for scoring that particular pair.
      Voting is simple - rank best liked candidate first, and continue 
down desires until remainder seem not worth bothering with.  Per above, 
candidates can be voted as tied in rank.
           Simple to do, for it asks nothing of the voter beyond ordering 
the candidates - something that is necessary even for Plurality voting.
           Little if any need for exotic schemes, or fear of others using 
such schemes to get an edge.
      Counting is simple and result for a precinct, or even a state, are s 
simple matrix, even for a dozen candidates.  Throw in 6 absentee votes a 
few days later and:
           Adds up to 6 to some of the counts - no surprises and no 
changes in winner unless previous counts were close to a tie as to winner.
           Avoid IRV's threat to possibly discard losers in a different 
order and, as a result, perhaps radically change the final result.

> 
> David Gamble

-- 
davek at clarityconnect.com  http://www.clarityconnect.com/webpages3/davek
  Dave Ketchum   108 Halstead Ave, Owego, NY  13827-1708   607-687-5026
            Do to no one what you would not want done to you.
                  If you want peace, work for justice.