[EM] Dave, IRV, 4/20/12

2012-04-20 Thread Michael Ossipoff
You said:

I choke when I see IRV called fine

[endquote]

Have I ever said that, without qualifying it? No.

I've said that IRV would be fine with an electorate different from
the one tht we now have--an electorate completely free of inclination
to overcompromise, so that even IRV's flagrant FBC failure wouldn't
induce them to overcompromise.

I've said that IRV would be fine for me, as a voter.

I'm not one of those who is inclined to overcompromise for a lesser-evil.

Its MMC compliance and defection-proofness would work fine for me.

You continued:

 - it too easily ignores parts of
what the voters say.  For example, look at what can happen with A
being much liked, yet IRV not always noticing:

20 A
20 BA
22 CA
Joe ?

Condorcet would see A elected by 62 votes (plus, perhaps, Joe's
63rd).

[endquote]

It would help to specify more about Joe. Examples with a voter whose preferences
and vote are unknown are difficult to comment on.

A and B voters are a majority, but not a mutual majority. the A voters
are indifferent
between B and C. So, maybe you're pointing out that for {A,B} to win
or not win, it depends on
which one gets eliminated first. True. Not ideal, I agree, but the B
voters want the coalition
and the A voters don't. So whether there's a coalition will depend on
which one gets eliminated
first.

And we do know that the A voters are indifferent between B and C,
because IRV gives them
no incentive to defect.

Mike Ossipoff


**

Election-Methods mailing list - see http://electorama.com/em for list info


Re: [EM] Dave, IRV, 4/20/12

2012-04-20 Thread Dave Ketchum

On Apr 20, 2012, at 5:30 PM, Michael Ossipoff wrote:
You said:

I choke when I see IRV called fine

[endquote]

Have I ever said that, without qualifying it? No.

I've said that IRV would be fine with an electorate different from

the one tht we now have--an electorate completely free of inclination
to overcompromise, so that even IRV's flagrant FBC failure wouldn't
induce them to overcompromise.

I've said that IRV would be fine for me, as a voter.


I'm not one of those who is inclined to overcompromise for a lesser- 
evil.


Its MMC compliance and defection-proofness would work fine for me.

You continued:

 - it too easily ignores parts of
what the voters say.  For example, look at what can happen with A
being much liked, yet IRV not always noticing:

20 A
20 BA
22 CA
Joe ?

Condorcet would see A elected by 62 votes (plus, perhaps, Joe's
63rd).

[endquote]

It would help to specify more about Joe. Examples with a voter whose  
preferences

and vote are unknown are difficult to comment on.
Mike chose to ignore the rest of what I wrote.  I will copy that at  
the end and comment.


A is well liked - except for Joe, every voter votes for A.

B and C contend, with NO voter voting for both.
A and B voters are a majority, but not a mutual majority. the A  
voters are indifferent
Huh!  B and C each got 1/3 of the votes - about tying each other, but  
far from a majority.
between B and C. So, maybe you're pointing out that for {A,B} to win  
or not win, it depends on
which one gets eliminated first. True. Not ideal, I agree, but the B  
voters want the coalition
and the A voters don't. So whether there's a coalition will depend  
on which one gets eliminated


first.

And we do know that the A voters are indifferent between B and C,  
because IRV gives them

no incentive to defect.

Mike Ossipoff

End of my email, that Mike did not include:


Condorcet would see A elected by 62 votes (plus, perhaps, Joe's  
63rd).  IRV would be affected by Joe's vote:

. A - 63 votes with B and C discarded.
. B - 22 for C after 20A and 21B20A discarded.
. C - 23 votes with A and B discarded.


Joe could have voted for A, B, or C, and have this noticed by IRV.  A  
vote for A or C would cause them to win; a vote for B would cause C to  
win.


DWK
Election-Methods mailing list - see http://electorama.com/em for list info