[EM] Condorcet How?

[robert bristow-johnson]

sorry if this is too long....

On Mar 21, 2010, at 7:45 AM, Markus Schulze wrote:

> Kristofer Munsterhjelm wrote (21 March 2010):
>
>> Schulze's advantage is that it's actually being
>> used and that it provides good results (by the
>> Minmax standard). Ranked Pairs's (or MAM's,
>> rather) is that it's easy to explain.
>>
>> The question is: which of these qualities are
>> more important, were we to encourage the use of a
>> Condorcet method in real (governmental) elections?
>
> I claim that Burlington repealed IRV mainly because
> it chose a candidate with a strong worst defeat.

the person elected was the 2nd most preferred candidate, not the  
least preferred by any measure.  do you mean that it's because the  
IRV winner's worst (and only) pairwise defeat was nearly 54% to 46%?

the cause of the Burlington repeal has a lot to do with local  
politics.  this mayor was the only mayor to be elected and re-elected  
with IRV and no other mayor was elected using IRV before, even though  
a couple of Progs (one is Bernie Sanders, now a U.S. Senator from  
Vermont) were elected from the old 40% plurality rules.  the  
political thing that really doesn't have anything to do with IRV is  
that since being re-elected, the mayor might have gotten into a  
little hot water regarding the city's credit and debt regarding a  
project called "Burlington Telecom".  the Republicans (and some  
others) always felt that this was an appropriate function for private  
commercial interests and not for government.  then when BT was being  
a money drain, if i understand it correctly, the mayor has used the  
cities credit to prop BT up a little.  this made him very unpopular  
very soon after his re-election (again with IRV) and then some of the  
blame went in the IRV direction.

now, even at the 2009 election, the Republicans were unhappy that  
their "lead" (plurality) in the election was held until the final IRV  
round and then lost to the Prog candidate (the incumbent mayor).   
this should not have been surprising, for this is why we adopted IRV  
in the first place.  why bother to change the election rules if you  
don't think it would ever make a difference from the old method?  but  
in 2006, the elected candidate (the present mayor) was elected with  
IRV and was also the Condorcet winner and the Plurality winner.  so  
there was little dispute then who should be mayor.  but i'll bet that  
no matter how a preferential ballot election was decided (IRV,  
Condorcet, whatever), as soon as it picked a different candidate than  
the plurality winner, *if* that plurality winner (who *loses* the  
election) was the Republican, they would have started bitching then.   
no matter how legitimate or non-anomalous the election was.

in fact, if the election was decided with Condorcet rules (doesn't  
matter which, since there was no cycle), these same Republicans would  
have bitched all the more, since the Condorcet candidate had only 23%  
and came third in plurality.  so the primary political motivation  
behind the repeal was not due to that the Condorcet winner was not  
elected.

but the fact that the Condorcet winner was not elected was used as  
grist for the mill by people who's political interest was far from  
those who would earnestly support Condorcet (Warren, are you  
listening?).  So the failure to picking the Condorcet winner was a  
criticism of IRV ("non-majoritarian"), but no one in Burlington  
(other than me) seemed to suggest that we keep the Preferential  
Ballot and reform the tabulation method from IRV to Condorcet.  the  
solution was just swept under the rug.

then the fact that the Condorcet winner was not elected with IRV  
caused a cascade of anomalies like Spoiler Lite (the Plurality winner  
was a non-independent irrelevant alternative), the punishing of a  
voting group (the GOP Prog-haters) for sincerely ranking their  
favorite candidate #1 (this leads to tactical voting in the next  
election if conditions are similar), and non-monotonicity.  since he  
was local, Prof. Anthony Gierzynsky's negative analysis (and polemic)  
of the 2009 IRV was oft cited by the opponents.  his "report" was oft  
held up as an academic indictment against IRV.

but most local Burlington voters didn't really understand nor care  
about any of those legitimate failures of IRV.  the big plug was an  
insinuation that the Republican plurality winner was the candidate  
robbed of the election and that he would have saved the city from all  
of that BT mess had he been elected. that was the essential root  
politics behind the repeal effort and the fig leaf they used to cover  
themselves was the "Keep Voting Simple" slogan.

my disappointment (and it was my unease when i voted for the adoption  
of IRV in 2005) was and is that nowhere is the ranked ballot ever  
introduced as a progression or reform to the "traditional" plurality  
(in a governmental election) without also being attached to the IRV  
method of tabulation.  in fact, many times FairVote and the local  
followers (like in Minnesota or Washington) literally *equate* the  
meanings of "Preferential voting" with IRV in their lit, and that has  
always been a big, big mistake and FairVote has yet to recognize  
that.  i still can't believe that Rob Richie still asserts that when  
IRV and Condorcet differ on the result (the Condorcet winner would  
have to be eliminated before the IRV final round), that IRV selects a  
"better" more democratically-preferred candidate.  that simply makes  
me incredulous. i just cannot identify with any democratic value  
system that says that *any* candidate should be elected with a  
majority of voters agree (and so mark their ballots) that some other  
candidate is a better choice.  the reason that Condorcet beats IRV or  
Borda or Range or Approval or Bucklin or the old traditional  
Plurality is fundamental: "If a majority of voters agree that  
Candidate A is a better choice than Candidate B, then if at all  
possible, Candidate B should not be elected."  (of course, it's  
unavoidable if there is a cycle.)

--

r b-j                  rbj at audioimagination.com

"Imagination is more important than knowledge."