Truncation Resistance #2 criterion

[Steve Eppley]

Demorep wrote:
>Mr. Eppley has the cart before the horse in his 46-10-10-34 examples.
>Whether or not a candidate is the compromise candidate in a
>circular tie depends on how the voters vote.  Saying that candidate
>X is the compromise candidate if the voters do not make him/her such
>compromise is somewhat illogical.  
-snip-

If Demorep re-reads the example, he will see that I wrote that B is
the compromise choice in the *left* side of the example, in which
there was no circular tie.

>Mr. Eppley is apparently assuming that (1) candidate A knows
>exactly how the 54 combined B and C voters will be voting and that
>his/her 46 voters are robots subject to A's total influence (which
>may or may not be true) and (2) candidates B and C know nothing
>about how the A voters are voting (or not voting) for their second
>choices and have no influence over their respective supporters.

How did Demorep reach that absurd conclusion about what I'm assuming??

>The general cases for first choices just as well might be the
>extreme cases--
>49 A, 3 B, 48 C
>or
>49 A, 25 B, 26 C.

If there's a point being made by that, I don't see it.  What does 
it mean for a "general case" to be an "extreme case"?

>As has been repeatedly pointed out, any election method (even
>Condorcet as amazing as that may be to some) has strategic voting
>possibilities.  For example, in some methods, the mere making of a
>second choice by a voter can defeat the voter's first choice (when
>combined with the first choices of other voters).  Example- the 34
>CB voters obviously help B become the compromise winner in the left
>46-10-10-34 example.

That's erroneous: C would *not* have won if the 34 CBA voters voted 
just C:

   46: ABC
   10: BAC
   10: BCA
   34: C

        A    B    C
   A        46L  56L
   B   20        66L
   C   44   34
      ---- ---- ----
   LL:  *   46   66   

By truncating, the C voters elect their greatest evil A, not C.  
So the C voters have a clear incentive to vote B their second choice.

>I must say again that if a single winner method does not guarantee
>a majority of all voters (moav) winner (51 of 100), then it will
>have almost zero chance of being adopted. 

No singlewinner method can guarantee there will be a candidate 
who receives a majority of all voters, when there are more than 
2 candidates.  So Demorep is proposing a criterion that no method 
can satisfy.  That's rather useless...

I'd like to see Demorep's polling data to know where he got that 
"almost zero chance" figure.

>Some of the competition to head to head remains approval voting, top
>2 runoff and instant run off which all will likely produce such a
>winner (but who may not be head to head compromise winner).

Demorep should understand by now that the so-called "majority" 
produced by 2-Runoff and IRO is bogus.  I believe he's acknowledged 
this already.

Smith//Condorcet and Condorcet are more about majority rule than
any of the "competition."  That's why they finished 1-2 in the 
EM poll.

>Thus, I note that if the first 2 choices are deemed acceptable to
>the voters in the left example 
-snip-

That's an amazing and irrational assumption; no such info is 
available from their ballots.  And such is impossible anyway, since
"approval" is really a restricted pairwise preference method--no 
absolute like or dislike is possible.

>Thus, I suggest that no method be posted on this list that does not
>guarantee a majority of all voters winner (i.e. a MOAV criterion). 

It seems Demorep wants us to stop posting any methods.  Perhaps he 
should more clearly define his MOAV criterion, if he thinks it's 
possible for some method to guarantee satisfaction.  Here's a little
test case for MOAV:
   33: A
   33: B
   34: C

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