Would Condorcet voters vote for only one? (was Re: Housekeeping

[Steve Eppley]

Donald D wrote:
> THREE: Why do I think the voters will only make one Condorcet
>        selection?
>
>QUESTION THREE Answer: I have claimed a number of times that in a
>Condorcet election most of the people will only make one selection -
>I still hold the same belief - belief is the key word here - I have
>no proof. "The proof of the pudding will be in the eating".

>From whence comes the belief in the pudding?

>     My logic says that even if Condorcet is voted in as the single
>seat election method of a voting area there will be people who did
>not vote for it. If these people decide to continue not to favor
>Condorcet they can do that by only making one selection. 

A pretty silly argument.  The same silliness could be said about any
voting method, including MPV and Coombs.  Donald should spend more
time looking at the voters' more plausible incentives: they vote
because they want to influence election outcomes.

>Can Condorcet survive if forty-five percent of the voters only make
>one selection? A Condorcet person should be able to answer this
>question better than I could. 

Well, I advocate Condorcet (and Smith//Condorcet).  But I can't
answer this unless Donald explains what he means by survival.

And it would be nice to know where the 45% figure came from.  Can MPV 
or Coombs "survive" if 99% of the voters make only one selection?  :-)

>But - my logic also tells me that additional voters will begin to
>only make one selection when they realize that their other
>selections are being used to help some other candidate succeed while
>their first selection is still a contender. I have stated this
>before.

***   ALERT!  DISORDERLY DISCUSSION!!   ***

Yes, Donald has stated this before and here he has repeated it
again.  The problem is, this has already been rebutted and Donald
has not attempted to rebut the rebuttals.  This is "disorderly 
discussion" and in my opinion is as bad for the maillist as flaming.
Now he's gone and done it again.  [Further flames deleted.]

Though it's tiresome to repeat the rebuttal, I'll do it briefly:
Ranking additional candidates after one's favorite helps defeat the 
even worse candidates, which is one of the goals of the voter.  It 
does *not* increase the "votes against" one's favorite.  Therefore, 
the clear incentive is for the voter to rank additional candidates,
because though it helps additional candidates succeed that won't be
in elections where their first choice would have otherwise won.

Here's a simple example:
   Suppose 45% of the voters have the preference order ABC.
   Suppose 10%               have the preference order BAC.
   Suppose 10%               have the preference order BCA.
   Suppose 35%               have the preference order CBA.

Donald suggests the votes (using Condorcet) will be something close to:
   45% A
   20% B
   35% C
   The winner with these votes would be A.
   (A pairwise-beats B, 45 to 20.  A pairwise-beats C, 45 to 35).

But Donald hasn't explained why the C voters would vote so inanely.
If the C voters vote their true preference order CBA, here's what 
would happen:
   45% A
   20% B
   35% CBA
   The winner with these votes would be B.
   (B pairwise-beats A, 55 to 45.  C pairwise-beats B, 35 to 20.
   (A pairwise-beats C, 45 to 35.  B wins the circular tie with a
   Condorcet score of 35.)

So the C voters do NOT cause C to lose by voting CBA instead of just
voting C.  They cause candidate A to lose!  Wake up, Donald.

They have a strong incentive to make A lose, since A is their last
choice.  Similarly, the other voters have strong incentives to rank
more than just their top choice.

Donald hasn't looked at the voters' real incentives, merely
postulated that many would be disgruntled and refuse to vote wisely.
He's claimed that ranking more than one candidate would hurt that
candidate, but he hasn't demonstrated this.  He continues to
demonstrate that he doesn't understand pairwise tallying and the
"votes against" Condorcet principle, which clearly make it wise for
voters to rank more than one candidate.  Maybe it's time he provides
us the long requested example showing how some voter will produce a
better outcome for him/herself (i.e., change the result so the
winner is a candidate s/he prefers more than the one who otherwise
would have won) by voting for only one candidate.  Until then, 
Donald should not repeat the rebutted conjecture. 

---Steve     (Steve Eppley    seppley at alumni.caltech.edu)