[EM] draft summary for Washington state legislation

James Green-Armytage jarmyta at antioch-college.edu
Sat Sep 10 14:53:11 PDT 2005


These are the recommendations that I sent to the new WA Condorcet list.
Hopefully I won't do too much double-posting, but I thought that this
message would make sense on both lists. 
________________

Hi there,
	I recently joined this list. In hopes that it will be helpful, I have
prepared a draft summary of interesting, uninteresting, and objectional
methods as I see them. These opinions are subject to change based on
further argument, of course. I think that it is helpful to provide this
kind of summary; ultimately, the legislators might want something with a
similar format, though significantly expanded. 

I. Interesting methods
A. In order of simplicity
1. Approval voting
2. IRV (preferably with equal rankings allowed)
3. CWO-IRV
4. Winning votes with SSD base (SSD here is Smith sequential dropping:
"Drop the weakest defeat in the top cycle until there is an unbeaten
candidate". I think equivalent to beatpath in public elections.)
5. DMC (Please explain with RAV heuristic, "eliminate the least approved
candidate until there is an unbeaten candidate". Simpler rule than
SSD(WV), but ballot harder to use) 
6. AWP with SSD base 

B. In order of merit
1. AWP
2. CWO-IRV
3, 4, 5, 6. very tentative ordering: 3. DMC, 4. SSD(WV), 5. approval
voting, 6. IRV

II. Methods that I don't consider to be interesting
1. Smith//approval. Not really a bad method, but no advantages over DMC.
In general, I would avoid inelegant methods that say "first try one thing,
and if it doesn't work, try something else totally different", as they
will seem awkward to the public, and may create perverse strategic
incentives.
2. Condorcet methods not passing Smith. My opinion is that Smith failure
is inconsistent with the basic logic of pairwise comparisons. If one
agrees that a candidate who beats all other candidates should win, it
seems to logically follow that if there is no CW, the winner should at
least come from the set of candidates such that every member beats every
non-member. To do otherwise seems to imply that the logic of Condorcet
breaks down when there is no CW. Also, methods like minimax have criteria
disadvantages to IRV (mutual majority, Condorcet loser, clone-proofness),
which is unnecessarily embarrassing to Condorcet methods.
3. Two round runoff. IRV seems to be clearly preferable to this. Although
it does require new ballots, this change to ranking ballots is IMO a good
thing in the long run as it opens up more possibilities. Otherwise, two
round runoff adds the extra cost of a second election, is marginally more
strategy prone than IRV, and does a somewhat worse job of solving the
vote-splitting problems found in plurality.

III. Methods that I consider to be objectionable
1. A margins-based Condorcet method. Link to my essay on the topic:
http://fc.antioch.edu/~james_green-armytage/vm/antimargins.htm
2. Borda. Link to my essay on the topic:
http://fc.antioch.edu/~james_green-armytage/vm/antiborda.htm
3. Bucklin. Hopefully this is not on the table.
4. Various others that are probably not under consideration

IV. "Constitutionality problems"
I'm not a lawyer, but here's my best advice on the topic: IMO, all of the
methods in (I) are constitutional, but in practice, something is
determined to be "unconstitutional" if and only if a judge doesn't like
it. Thus, any of the above methods (with the exception of two round
runoff) could be considered unconstitutional depending on the opinions and
prejudices of particular individuals. IMO, the most likely to survive a
challenge would be IRV. The least likely to survive a challenge would be
CWO-IRV. The rest would face a significant challenge as well. E.g.
approval advocates will need good lawyers and a relatively broad-minded
judge to prevail, because the "one person, one vote" dictum can be used as
a justification for striking it down. 

all my best,
James




More information about the Election-Methods mailing list