51/48 example

[Steve Eppley]

Demorep wrote:
>A divided majority and divided minority example--
>
>99 voters
>51 voters in the ABC group
>48 voters in the XYZ group
>Votes to the left of the slash (/) are approval votes and votes to the right
>of the slash are disapproval votes.
>17 ABC/XYZ
>17 BCA/YZX
>15 CAB/ZXY
>2  CAB/ZYX
>16 XYZ/ABC
>16 YZX/BCA
>15 ZXY/CAB
>1   ZXY/CBA
-snip-
>With plain Condorcet, Y (with majority disapproval) wins with a
>mere 64 votes against him/her in his/her worst defeat.       
-snip-

In Smith-Condorcet, since X, Y, and Z are all pairwise-defeated
by A, B, and C, the winner is chosen from among A, B, and C.

>Obviously the example is extreme (a majority group circular tie and
>a minority group circular tie) but not impossible. Note that if the
>XYZ variants are dropped, then A, B and C each have 33 first choice
>votes.

The majority approval filter is a cousin of the Smith filter, but
Smith-Condorcet won't leave a vacuum of power so it would be less
controversial.

Is it worth worrying about extreme examples?  The replacement of 
candidate names (like Hitler, etc.) with neutral letters obscures the 
issue of whether the example is politically realistic, seems to me.

>In a small election, there is obviously a much higher possibility
>of ties. How many voters are necessary for a "large" election ?
>Answer- the method used should not matter regardless of how few or
>many voters there are.

Is is wise to talk about the tie-breaker portion of the method before
people understand the main portion, if that's all that would be used
99.9999999% of the time?  By the time a tie actually occurs in a 
large election, all the people who we've taught the tie-breaker will 
be deceased.

If a tie occurs, does it really matter much how the tie is broken, 
as long as basic democratic principles won't be violated?  It's 
not nearly as important as getting the basic method done right.  
I wouldn't oppose Condorcet//Random if that's what a group desired, 
but Condorcet//Plurality-ext is simple and practical. 

Maybe the tie-breaker should be in the fine print, or left unchanged
by the sw reform.

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