[EM] Real IRV Election, Disputable Result

[Jonathan Lundell]

At 12:47 PM -0500 3/14/06, Dave Ketchum wrote:
>SUMMARY of what I see since Rep. Jim Condon asked for help:
>
>With IRV, tie-rank votes need to either be prohibited or the exact way of
>accounting for them defined (counting them with each tied candidate
>holding the same rank encourages taking advantage of such voting).  With
>Condorcet, tie-rank should be normal, for voters can like it and it gives
>no special advantage to users with the round-robin counting used.

As a point of clarification, the Burlington rules basically don't 
allow equal-preference ranking. However, they do try to salvage a 
ballot so marked if it can be done unambiguously. The relevant rule:

>4. A ballot that gives two or more candidates the same ranking is 
>exhausted when that ranking is reached, unless only one of the 
>candidates so ranked has advanced to the current round of counting.

(Also, if all the candidates so ranked have been eliminated, the 
ranking is skipped as usual.)

I think it's a reasonable way to handle equal ranks if the voting 
equipment can't exclude them at voting time.

>On Tue, 14 Mar 2006 00:55:02 -0800 Brian Olson wrote, and I suggest reading:
>
>He mentions difficulty with combining multiple precincts in a district.
>My response is that, while it is bad, it is not as bad as he suggests -
>you can note that multiple voters can vote identical patterns and pass up
>counts for each pattern for central combining:
>       Besides counts of nulls and spoiled ballots (I here count equals as
>spoiled with IRV), and counting votes for all but one candidate the same
>as listing that last candidate as least liked:
>       2 candidates - same 2 counts as with Plurality.
>       3 candidates - A  B  C  A,B  A,C  B,A  B,C  C,A  C,B
>       4 candidates - A  B  C  D  A,B  A,C  A,D  B,A  B,C  B,D  C,A  C,B
>C,D  D,A  D,B  D,C  A,B,C  A,B,D  A,C,B  A,C,D  A,D,B  A,D,C (and 18 more
>with each other candidate as leader)
>       more - number of possible patterns goes up, but usage of all becomes
>less likely

Jeff O'Neill (who I think is on this list) suggests a tree 
representation of an ranked election profile (Voting matters #21 
<http://www.mcdougall.org.uk/VM/ISSUE21/INDEX.HTM>) as a means of 
speeding up the tabulation of the election. It seems to me that such 
a representation, suitable represented in a text file, could be used 
to collect and aggregate ballots from district subdivisions.

>On Mon, 13 Mar 2006 21:53:44 -0800 Jonathan Lundell wrote, and I see
>nothing valuable to read.

I get that a lot. :-(

>On Mon, 13 Mar 2006 22:58:36 -0500 Eric Gorr quoted an article worth
>reading by Ralph Suter.

I suggest that James Green-Armytage's reply to CVD's defense of IRV 
is a better treatment of IRV-vs-Condorcet issues. 
<http://fc.antioch.edu/~james_green-armytage/vm/cvdletter.htm>

JG-A does an evenhanded job of balancing CVD's rather tendentiously 
pro-IRV piece, without ignoring the serious problem that Condorcet 
methods have with strategic "burying". Recommended, along with the 
linked piece that tries to address this problem (with, I'd say, mixed 
success).

We can all agree that if 1) all voters sincerely rank their 
preferences, and 2) those rankings result in a Condorcet winner, that 
the Condorcet winner should be elected. Suter acknowledges (2) but 
entirely ignores (1).
-- 
/Jonathan Lundell.