Hi all,
On Jun 1, 2004, at 3:38 PM, Dave Ketchum wrote:
Recent French elections demonstrated need for something better than Plurality plus rerun. I believe they also demonstrated that IRV does not cut it - IRV too easily locks out acceptable candidates when minorities each rate a few minor candidates as best.
To me this leads to Condorcet, using computer-based voting and counting. We should advise careful use of Condorcet in a single pass election - that it needs neither primaries nor reruns.
Someone sent me the following off list (and asked it be forwarded anonymously):
1. Approval
This has been suggested elsewhere (can't find the reference at the moment
...). Use Approval Voting on the first ballot, cut off the top N
highest-approved candidates. The suggestion I've seen is the top 5
candidates, then use Condorcet for the general election.
See this post, it probably qualifies as the first mention, and includes a nice
example of usage:
http://lists.electorama.com/pipermail/election-methods-electorama.com/ 2001-May/006053.htm
Incidentally, if the approval cutoff is taken with the top 3 candidates, and
winning votes (not margins) are used, there is essentially no difference
between Ranked Pairs and Beatpath. Isn't that attractive?
Top 3 approval allows for the possibility of (say) D & R candidates and one
strong outsider, or 2 strong candidates from one major party and one from the
other. So 3 candidates are not ideal in terms of eliminating the clone
problem, but in practice, if two of the three strongest approval candidates
are in one party, that is a pretty good indication that the electorate is
leaning that way anyway.
2. STV-PR
the most recent post to election-methods suggested PR as appropriate for a
multi-seat election.
I like that idea! Except I would use 3 seats and then Condorcet for the
general election.
This might be a good idea for open primaries as well -- the question is
whether it removes the incentive for members of one party to strategically
misrank an unelectable opposition party member higher than the more electable
candidate. Hmm ...
Okay, on investigating further, I see a better alternative.
Say there are 3 "seats" for the general election, which will be decided by wv
Condorcet (RP essentially the same as BP in this case).
Single Transferable Proportional Vote is the best way to select them in a PR
scheme.
Why? It's because STPV, unlike STV, doesn't have a threshold for winning a
seat, exactly the criterion we want for a primary election. It just
eliminates the lowest first-ranked candidate and redistributes votes. This
process repeats until 3 candidates remain. This method also wastes the fewest
votes.
STPV reduces tactical voting in the primary since one places one's
most-preferred candidates highest to prevent their elimination. It is also
unlikely that, say, a Republican voter would rank a unelectable Democratic
candidate higher than his/her own favored candidate.
So, I interpret that -- if one assumes the existence of a primary -- as recommending STPV-PR for the first round, then a three-candidate Condorcet for the final round.
In the California case, if we can't get Condorcet for the general election, we'd have to pick two candidates, and it still sounds like either Approval or STPV-PR would be the optimal primary.
Thoughts?
-- Ernie P.
---- Election-methods mailing list - see http://electorama.com/em for list info
