[EM] More Ballots

[Eric Gorr]

I was searching through USENET looking for references to Condorcet 
and came across the FSCC Competition in rec.puzzles.crosswords. 
Apparently they run a contest where people create 5x5 crossword 
puzzles and then others vote on their favorites.

In any case, was able to find ballots for 33 out of the 41 completed 
competitions in the USENET archives.

In 22 of those contests there was a Condorcet Winner(CW) and IRV only 
selected it 15 times. 6 times it may have selected the CW depending 
on how the random eliminations went (see: http://demochoice.org, 
sponsored by the Center for Voting and Democracy). In one case, it 
simply would not select the CW.

If you count all 7 cases as failures, IRV failed ~31% of the time.
If you count just the one case where IRV would not select the CW, it 
is a ~3% chance of failure.

Someone said in the past that IRV and a good Condorcet method will 
only disagree on rare occasions. Even at 3%, I do not consider that 
to be very rare and this is quite consistent with others ballot sets 
I've tested.

You can the ballots at:

http://www.ericgorr.net/library/tiki-index.php?page=FSCCArchives

The following USENET google search will turn up the archived ballots:

http://tinyurl.com/2cvv2