[EM] About IRV

[Dave Ketchum]

On Thu, 22 Apr 2004 11:40:53 -0400 (EDT) wclark at xoom.org wrote:

>>Thought you all might be interested in this
>>
>>http://www.utahpolitics.org/archives/000173.shtml
>>
> 
> That's a pretty scary article.  Most of those claims don't just apply to
> IRV, but to pretty much any ranked system other than plurality.  It's
> pretty much a pro-status-quo piece, more than anything else.
>
Following posted in the archives just now.
---------------------------------------------

Dana Dickson's article says "IRV IS BAD" in various ways.  As others have 
noted, many of his arguments are weak.

For example, he talks of IRV resulting in enough candidates to cause a 
variety of problems.
---Among many methods, including Plurality, if the method of doing 
nominations permits such an excess, you can get there - look at the recent 
CA recall election.  New York State standard is to require a wannabe 
candidate to get approval from 5% of the voters - a theoretical limit of 
20 candidates; a lower practical limit since each voter can sign only one 
petition and wannabe candidates have to find these voters.
---Ranked ballots, letting voters indicate support for minor candidates, 
while ALSO indicating support among major candidates, could inspire more 
minor candidates, BUT not to the extent Dana suggests unless one of 
these minor candidates earns major support.

My purpose here is to note that IRV and Condorcet (tournament) are VERY 
similar, yet Condorcet is enough better that going to Condorcet in one 
step is better than going to IRV, getting exposed to its problems, and 
then having to battle to get another change approved.
---IRV simulates instant runoffs.
---Condorcet looks at complete ballots, noting relative liking among each 
pair of candidates.  Note that the array of comparisons is useful data, 
aside from its immediate use in the election.

What I see:

Both use identical ranked ballots.  Condorcet CAN permit ranking 
candidates as equally liked - not clear to me whether this is practical 
for IRV.

Voting rules are simple - rank the candidates you like best in order, and 
ignore those you like too little to care about.

For MOST elections IRV and Condorcet will agree as to winner, although 
their methods of counting differ.  Major cause for disagreement is three 
or more candidates about equally liked, and IRV responding by ignoring 
parts of ballots that matter, and selecting a non-deserving winner.

IRV offers strategy possibilities for those willing to work at this.  The 
work is to decipher how your enemies will vote, and then have your friends 
respond to this prediction rather than voting their desires:
---Backfires if your enemies play the game smarter.
---Existence of such possibilities annoys everyone not playing the game.

Condorcet can notice near-ties - often called cycles (e.g., A>B & B>C & 
C>A).  Often made out to be a bigger problem than it is.
---Since they are near-ties, all that is needed is a reasonable rule for 
resolution,

Approval got mentioned.  While this method sounds simple, how do I vote 
among LIKE, who I obviously approve, HATE, who I obviously skip, and COMP, 
  who I want to show as preferable to HATE, but NOT to aid COMP possibly 
beating LIKE?
-- 
  davek at clarityconnect.com    people.clarityconnect.com/webpages3/davek
  Dave Ketchum   108 Halstead Ave, Owego, NY  13827-1708   607-687-5026
            Do to no one what you would not want done to you.
                  If you want peace, work for justice.