Majority Tie Breaker

[Steve Eppley]

DEMOREP1 wrote:
-snip-
>The plain Condorcet method will often fail to produce a winner.

Why do you say that?  With a large number of voters, the odds of two 
candidates tying with an identical smallest worst "votes for opponent 
in lost pairings" are nearly as negligible as the odds of plurality 
producing a tie.

On second reading, though, I guess you are saying that with plain
ranked ballot methods there will often be no candidate who beats
each of the others pairwise, or that there will often be no candidate
who has a majority of the first-rank votes.  If these are
requirements of winning, then we're no longer talking about plain
Condorcet--we'd be talking about the Beats_All method or the
Majority_of_First_Ranks method (which are both often indecisive).

(By the way, if there is a winner using Beats_All or Majority_of_
First_Ranks, then the same ballots tallied by Condorcet's method will
produce the same winner.)

I repeat my request that you use the term "pairwise method" instead 
of "Condorcet" unless you're referring to the pairwise methods which 
give the win to the candidate with the smallest worst "votes against."
It's true that the term "Condorcet winner" is used by academics to 
mean the Beats-All candidate, if there is one (although there's some 
ambiguity over whether they mean sincere beats-all or voted beats-all), 
and it's also true that the Marquis de Condorcet deserves some credit
for his contributions to this field of study.  But we can make our
communications less ambiguous if we reserve the term Condorcet for
the "smallest worst pairdefeat" methods, and use general terms like
"pairwise method", "sincere-beats-all-pairwise", "beats-all-pairwise", 
etc., where appropriate.

-snip- 

>Thus, there is in order--- 
>Plurality 
>Condorcet 
>Approval

If there's a candidate who has a *sincere* majority of first choice 
votes, this candidate will win if Condorcet is the method.  So to me 
it seems redundant to put plurality first if you're requiring a 
majority to win.

>The reason for a majority winner should be rather obvious. Any
>reform merely raising the minority percentage is hardly good enough
>(such as changing a 38 percent plurality winner to a 45 percent
>winner with *the* utopian single winner method). 

But having no winner is hardly good enough either, in many
situations.  

(If you're referring to reforms which transfer powers away from
single winners to proportional legislatures so single-winner
elections are less important, I certainly agree with that long-term
reform goal.  It's probably pushing beyond the scope of this maillist
to spend time on it, though.  But here's my pet plan, anyway: I'd
structure the government into "circles" ranging in size from one
person to the entire citizenry.  Small circles have an advantage of
speed so they can act quickly when time is of the essence.  But
larger circles would have more authority, so they can act when they
wish and undo decisions made by smaller circles.  Circles would be
able to create smaller circles for limited or general purposes, elect
the members of the smaller circles using a proportional method, and
they'd be able to terminate a smaller circle anytime.  The only
exception to the requirement for using a proportional method would
be for the circles of one member--the emergency-handlers--which would 
be elected using the best single-winner method.)

If the people are divided, the people are divided--I don't expect
the voting method to manufacture consent.  Any election method will
be adequate in the case where a majority of the voters have the same
first choice.  We're trying to find which method(s) are good at
electing the "best compromise" for the other cases. 

I don't have a strong objection to combining ranked ballots with
approval ballots (such as with the NOTB option), although I'm
concerned that the particular method used to tally these ballots 
may introduce unforeseen incentives to misrepresent one's preferences.  

Also, a capricious electorate might vote to approve only their first
choices, which would lead to frequent new elections and possibly
unending indecisiveness.  It might be better to trust that with a
good ballot method (like Condorcet, imho), enough decent candidates
will finally be able to run--unhindered by the spoiler dilemma (the
flip side of the lesser of evils dilemma)--so there won't be much
worry about a majority-disapproved candidate winning.  Won't good
candidates flush out the bad? 

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