Thinking out Loud

[DEMOREP1 at aol.com]

Forest wrote in part-

Use a Proportional Representation method in each round of a runoff to see
which subset of candidates survives to the next round.

The motivation for this idea is elimination of primaries.

----
D- Head to head for the single winner case is the limiting case of the 
multiple winners case ---

ALL combinations of 

Test Winners   versus Test Loser  (Other choices deemed Other Test Losers)

Votes from Other Test Losers go to a Test Winner or Test Loser.

If a Test Winner wins in all of his/her combinations, then he/she is a p.r. 
Condorcet Winner.

Standard problem -- circular ties are possible (i.e. not enough p.r. 
Condorcet winners).  A tiebreaker will generally be needed.

General tables--

1. YES (desired, compromise) / NO (unacceptable) table (with possible 100 to 
0 percentage scale votes)
2. Head to head table (the above)
3. Place votes table

Note -- because of the mere existance of such 3 types of tables there are 
various methods (with their various possible defects).

A simple multiple runoff method (with lots of time) might be just repeatedly 
vote NO on 1 choice (or remaining choice) --- plurality NO eliminations.

Choices with small support and clones would be eliminated sooner or later.