[EM] the name of the rose

[Dave Ketchum]

On Mon, 04 Aug 2003 10:46:31 -0400 James Green-Armytage wrote in part:

> Dear John Hodges, and other voting fans,
> 
> you write:
> 
>>Condorcet advocates might consider a name change; call it "tournament 
>>runoff". At least some types of sports (Chess, Judo, maybe others) 
>>hold tournaments where every contestant is paired off versus every 
>>other in turn. It might be easier to sell the idea to the public if 
>>you present it as a round-robin tournament between all candidates. If 
>>one emerges undefeated from all possible two-person runoffs, they are 
>>declared the tournament champion. If nobody is undefeated, then we 
>>name ________________ as the tournament winner.
>>
> 
> I reply:
> 
> 	Yes, I've been thinking about this. Obviously the name "Condorcet's
> method" is woefully inadequate for any kind of slick public relations
> campaign in the US.
> 	The term "tournament voting" occurred to me. "Tournament runoff" may
> indeed make more sense than that.
> 	Another possibility is "pairwise voting," a term which is already used to
> some degree in theory circles, is somewhat descriptive of how the method
> works, and which has kind of a nice ring to it, also happening to contain
> the word 'wise,' which is a good association to make with Condorcet's
> method.


It is not too early to be considering a salable name, and of what I see 
above I like "tournament voting" best, as it has meaningful content - each 
of the pairs is playing a round in front of each voter.  By ranking the 
candidates the voter is declaring round winners (and declaring ties among 
candidates ranked identically, and among the candidates left unranked via 
truncation).  Candidate winning the most rounds is winner within that pair.
      BTW - I say above that a voter can assign identical ranks when 
multiple candidates are seen as equally deserving - I like that and see it 
as both easily implementable and easily understandable (if X and Y are 
ranked together by two voters, each gets credited with one win - unlike 
truncated candidates who get no credit).
      If one candidate is winner against each other candidate, that 
candidate wins the election.
      Otherwise we have a near tie among the three or more candidates that 
sometimes beat each other but do all beat all candidates outside the 
group.  Here we drop the pair race that came nearest a tie and pick the 
election winner according to the other pair races.
      BTW -I do not use the word "runoff", and think it is best left out, 
even though cycle resolution could be called a runoff.  We are determining 
election winner from analyzing ALL that the voters say in a single 
election, rather than simulating a runoff as IRV does (and rather than 
ignoring part of what the voters say, as IRV often does).

"Condorcet's method"?  It still is, and should be referenced, POSITIVELY, 
as such in any document with room for such detail.

"Pairwise voting"?  Partly it does not say as much to me; partly I 
understand there are other methods based on pairs (probably other methods 
that could claim the label "tournament" - can we be first to seriously 
grab that label?

> 
>>The blank must be filled in a way that is easily understood. Suppose 
>>we said "the one who wins the greatest number of two-person runoffs". 
>>
> 
> 	I wouldn't want to compromise the completion method to make it more
> sellable. I would probably just fill in the blank by saying that the
> winner of a cycle is the one whose worst defeat is by the fewest number of
> points (Condorcet's own suggestion, equivalent to "sequential dropping,"
> or "successive reversal"), and then if anyone gets really interested (and
> when laws are being drafted and software written), I will add that the
> method should also be Schwartz-efficient and cloneproof. (Which would
> bring you to beatpath / CSSD... I'm not totally sure how you would sell
> ranked pairs in this way, but I'm sure it can be done if it turns out to
> be better.)


I tried to cover this above.  One detail I feel certain of - ALL WE HAVE 
for analysis is the numbers in the pairs matrix (which could be for a 
single precinct for a local election; for a whole state to elect a governor).

Some of the words I read in the above paragraph are proper for an election 
methods discussion, but seem of little value in selling to voters.

> 
> 	Anyway, trying to find a good name for a voting system is part of the
> process of advocacy. CVD is avid about renaming voting systems. Aside from
> christening "IRV," they have renamed proportional representation as "full
> representation," STV as "choice voting," and perhaps most ambiguously,
> they have renamed single non-transferable vote as "the one-vote system."
> 	I'm not saying that all of this isn't a little overboard, but I will
> stand by my earlier point that "Condorcet's method" is really a bad name
> for advocacy in the US, being 1. difficult to pronounce, 2. distinctly
> foreign and 'non-American,' 3. the name of a person (albeit a long-dead
> one), which somehow makes it seem a little bit too personal and egotistic,
> not something universal enough to select a president or governor, and 4. a
> name which says nothing about how the system actually works.
> 	I would be curious to know if anyone else has some ideas for slick names
> for Condorcet. At this point "pairwise voting" seems like a pretty safe
> bet to me, but I could probably be talked out of that.
> 
> 
> my best,
> James

-- 
davek at clarityconnect.com  http://www.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.