[EM] RE: Election-methods digest, Vol 1 #397 - 6 msgs

[MIKE OSSIPOFF]

Markus said:

In the voting recommendation to the DEBIAN project, you can find a very
interesting comment to this question. Norman Petry wrote (6 Feb 2001):

>Regardless of what names we use when referring to these methods during our
>committee discussions, I think it is appropriate that if one of these
>variants is recommended to Debian that it be named SCHULZE'S METHOD.  This
>is because:
>
>1. Schulze, version 1: The 'Beat-Or-Tie-path' method was first proposed by
>Markus Schulze on Sat, 4 Oct 1997 (see EM Archives, "Re: Condorect 
>sub-cycle
>rule").  Unfortunately, eGroups has not archived this message, but it can 
>be
>found at Rob's site, in this text file (but mistakenly referred to there as
>"Tideman's Method"): http://www.eskimo.com/~robla/em/archive/em.97q4

Norm, in that message, points out that "Schulze's method" uses 
beat-or-tie-paths rather than beatpaths. There's been some vagueness about 
which of those you refer to as "Schulze's method".

>
>2. Schulze, version 2: The 'Schwartz Sequential Dropping' (SSD) method was
>first proposed by Markus Schulze on Mon, 10 Aug 1998.  His description can
>be found here:
>http://groups.yahoo.com/group/election-methods-list/message/673

In that proposal, it isn't clear what you mean by "potential wininers". You 
don't define the term there. I suppose that if something is defined slopily 
or vaguely enough, that leaves much latitude for saying what it is.

No one can prove what you meant in that proposal.

In any case, I didn't mean to imply that I consider it important who first 
proposed SSD, or whether your proposal in '98 was SSD, CSSD, or neither.

>Markus' version of 'Cloneproof SSD' uses a slightly different (and
>imo better) tiebreaker than Mike's version, but the two methods are
>otherwise identical.

You'd suggested comparing margins when 2 opposing beatpaths (or 
beat-&-tie-paths?) are equal because their weakest defeat is equal. I hadn't 
bothered to add that, because I'm mostly interested in public elections, and 
because even in committees, equal opposing beatpaths won't be common. Of 
course there's nothing wrong with adding that provision.

There were aspects of your tie solution that you later retracted when you 
discovered that they had undesirable properties.

Markus said:

Another interesting question is: In so far as you considered neither
independence of clones nor reversal symmetry to be important, why
did you propose an election method that is more complicated than
Schwartz//MinMax(VotesAgainst)?

I reply:

I don't know what you mean by MinMax. MinMax has been used with so many 
meanings that it's quite useless as a method name. Perhaps you're using 
MinMax to refer to PC.

I agree that I don't consider reversal symmetry important. As for clone 
independence, it's desirable, but far from essential. But what I was saying 
in my recent message was that the clone independence _of a tiebreaker_ isn't 
important, because no one is going to have a strategy dilemma about what to 
do in order to take advantage of a tie. Ties of all kinds are vanishingly 
rare in public elections (unless Copeland is used). Even in committees, ties 
aren't common, and clone-independence of a tie solution isn't really 
important.

So far as I'm aware, SSD is clone-independent in public elections (where 
there are no pairwise ties), but not in coimmitees with few voters (where 
there might be pairwise ties).

MAM (Maximize Affirmed Majorities), also called Ranked-Pairs, is 
clone-independent in public elections and in committees.

Though clonen-independence is desirable, that isn't the main advantage, for 
me, that SSD and MAM have over PC. The more important advantages, for me, 
are:

1. PC fails Condorcet Loser, Majority Loser, and Mutual Majorilty (in 
descending order of likely importance in campaigns). Those criterion 
failures are important only because they could be used against a PC 
proposal. When those criteria are defined so as to meaningfully apply to all 
methods, Plurality fails them too, which of course greatly reduces their 
importance in campaigns to replace Plurality with PC. Still, possible 
criticism of PC in campaigns, by academics and IRV advocates is probably 
PC's main disadvantage with respect to SSD and MAM.

2. SSD and MAM have been shown to meet GSFC and SDSC. PC hasn't been shown 
to meet those criteria. If anyone can show that PC passes or fails those 
critreria, they're invited to do so.

I consider SFC and GSFC to be the biggest, and most exclusive advantages of 
Condorcet wv. The fact that SSD and MAM have been shown to meet GSFC 
greately increases their value.

GSFC is the deluxe SFC that applies even when there's no CW.

PC has been shown to meet SFC.

Mike Ossipoff

_________________________________________________________________
Working moms: Find helpful tips here on managing kids, home, work —  and 
yourself.   http://special.msn.com/msnbc/workingmom.armx