[EM] just to let you know ...

[robert bristow-johnson]

On Jan 6, 2010, at 6:49 PM, Terry Bouricius wrote:

> Actually, the opposition to IRV in Burlington is predominantly  
> focused on
> the complaint that the plurality leader in the initial tally ended up
> losing in the runoff tally.

which we should identify as an unjustified complaint.  the whole  
point is that someone could be the plurality winner but a majority of  
the electorate are united supporting another candidate against this  
plurality winner.  why (outside of a condorcet cycle) should we ever  
elect candidate B, when a majority of the voting constituency agree  
that candidate A is better than candidate B?  this is the kind of  
decision we would make if it were just these two candidates running.   
why should it be different (reversed) with the introduction of a  
third candidate C?  it's inherent to Condorcet to say that the number  
of voters who say A>B is not a function of C.  if A is ranked higher  
than B for a specific voter, it remains so no matter where candidate  
C is ranked relative to A or B.

the reason why Kurt Wright lost to Bob Kiss is entirely that more  
Burlington voters marked their ballots that they preferred Bob over  
Kurt for mayor.  those two facts are equivalent in IRV.  but for the  
final round only.  Condorcet asks that question for any hypothetical  
two-candidate elections.

> This candidate was actually the Condorcet
> LOSER among the top three candidates (though a fringe candidate  
> with only
> 35 votes was the technical Condorcet loser).

yeah, but if you don't count Homer "James" Simpson, the only  
insignificant candidate (the four other candidates were all  
credible), then independent Dan Smith was the Condorcet loser.   
actually, all five candidates were quite unambiguously ordered, in a  
Condorcet manner, as M>K>W>Sm>Si.  you could remove any set of  
candidates and the remaining relative ordering would be unchanged.   
at least for the 2009 Burlington election with ranked-order ballots.  
that just seems completely insensitive to irrelevant alternatives.

> The complaint from those
> circulating the IRV repeal petition is that there shouldn't be any  
> ranked
> ballots, and that the plurality winner with 33% of the vote in the  
> first
> round (and the essential Condorcet-loser) should have been declared
> elected.

how can we say that, Terry, when what result they are calling for in  
the repeal petition is to return to the previous law (40%+ or delayed  
runoff)?  if the old law applied and people's first IRV choice was  
their single candidate vote, then the runoff would be between Bob and  
Kurt.  assuming the turnout and relative preferences are identical,  
Bob beats Kurt.  but their hope is that more of them show up for the  
runoff than their lazy librul opponents.  that's the only hope they  
have of electing a GOP mayor in a town of liberal Democrats and Progs.

> There is no momentum toward a Condorcet approach currently.

i agree with that.  but i think it's dumb.  one side or the other  
will lose on the second day in March.  either way, that side that  
loses will prefer Condorcet over what they get as losers.  if they  
think about it.  the pro-IRV people would prefer Condorcet over  
pre-2005 and the anti-IRV people *should* prefer Condorcet over IRV  
because of what IRV did to them in 2009.  the IRV in 2009 penalized  
Wright voters, more of which preferred Montroll than Kiss by 3 to 1,  
for not forsaking their favorite candidate.  we know that if a few  
hundred of those W>M>K voters had stayed home, they would have gotten  
M instead of K.  they actually helped elect the candidate they least  
preferred by voting sincerely for their favorite candidate.  that  
rewards the strategy of "compromising".  that is the principle  
strategy we were seeking to be relieved of in adopting IRV.  we  
wanted to be able to vote for Nader, as a political statement, yet  
still not support Bush by doing so (or you could say the same for  
Perot and Clinton, respectively).  this is why i think it's dumb.

> I
> haven't heard more than a couple of people in Burlington suggest  
> that the
> actual Condorcet winner should have won,

there's a lot of people (587 more than think otherwise) that think  
that Andy Montroll should have won over Bob Kiss.  that's what i  
heard in March 2009 from how 7541 voters marked their ballots (84% of  
7984 cast).  i mean, Terry, if we respect the authority of the ballot  
(which, as an aside, the Supreme Court didn't 12/12/2000) isn't it  
essentially a tautology that the people are asked if they prefer the  
Condorcet candidate to any of the other candidates, and that this is  
more than a couple of people, but the will of the people.  isn't the  
authority of the ballot an axiom for reflecting the will of the  
people?  and if more people people prefer candidate A over candidate  
B then candidate B should not be elected, no?  isn't that what  
Condorcet essentially says?

just as symbols, let's use some common names for candidates A, B, C,  
and D.  let's say candidate A is, say, "Andy", and candidate B is say  
"Bob".  candidate C is, hmmmm, "Curtis" and candidate D is "Dan".  no  
matter who else is in the race, it's always A>B>C>D; Andy is always  
preferred over Bob and everyone Bob is preferred to.  Bob is  
preferred to Curtis and everyone else Curtis is preferred to, which  
is Dan.  that's how those ballots were marked for candidates A, B, C,  
and D in 2009.

with Condorcet neither candidate D nor candidate C affect the  
relative preference of candidates A and B.  but with IRV, with  
"Curtis" in the race, that reverses the outcome of Andy over Bob.   
Curtis is qualitatively a spoiler, he does not win and changes the  
IRV election between Andy and Bob.  but he doesn't with Condorcet.

and, it's appears likely that candidate Bob is the most-disliked  
candidate of those who most liked candidate Curtis.  those voters  
might begin to think twice about forsaking their sincere favorite  
candidate in order to not help elect the candidate they dislike the  
most.

> because he was a weak Condorcet
> compromise in third place in the initial tally. I suspect that if
> Burlington had used Condorcet rules and the candidate in third  
> place in
> the initial tally had been declared elected, there would be even more
> vociferous calls for repeal in favor of plurality or runoffs.

maybe you're right about that, Terry.  but it would be dumb.  i just  
have to sigh.  the Dems wouldn't be complaining about the outcome,  
the Progs would be relieved that it isn't Kurt, and the GOP  
complainers *should* be relieved that it wasn't the Prog they so  
hate.  that's what Condorcet would have done for those partisan  
groups if it was the law instead of IRV and the same voters marked  
identical ballots the same as they had last March.  now that's the  
political angle.  Condorcet favors the centrist candidate.

whereas IRV favors the larger subgroup (Progs vs. Dems) of the larger  
group (liberals vs. conservatives in Burlington).  unlike upstate NY  
where the Convervative-party candidate for US Congress was different  
than the Republican and the conservatives were split, in Burlington  
is the liberals who are a little more and split.

but whether or not an election method favors the majority of the  
majority group or favors the people around the median is no reason to  
adopt an election method.  The reason that the Condorcet winner  
should be elected is the same reason that the Majority candidate is  
elected in a simple two-person race; that when asked, the majority of  
voters said they preferred this candidate of the two.  If a majority  
of voters agree that candidate A is better than candidate B, then  
candidate B should not be elected.  That (majority rule) is the only  
reason and that is good enough.

--

r b-j                  rbj at audioimagination.com

"Imagination is more important than knowledge."