[EM] MSC reply

[Rob Lanphier]

On Wed, 28 Feb 1996 DEMOREP1 at aol.com wrote:
> B. With just 3 candidates (with 2 of them being "extremists") a "middle"
> voter has some elementary strategic voting to do for his/her second choice
> (who might be a minus 99.9 percent choice versus a minus 100 percent choice)
> as shown in my posting of 

They don't have to rank anyone as their second choice, unless they prefer 
one extremist to the other.

> C. The test winner, test loser and test other losers terminology is the same
> as what the Condorcet method does in making pair comparisons- one of the pair
> is a test winner, one is a test loser. The first choice votes for each of the
> pair and the votes transferred from the other first choice candidates (test
> other losers) for one of the pair determines which of the pair is a relative
> winner. Saying that the Condorcet method compares the votes each of two
> candidates receives as if there were only those two candidates may be good
> enough for speaking to folks but not for math examples on paper and election
> computer programmers.

*I* am a computer programmer.  I wrote a program that calculates the 
Condorcet winner.  I graduated with a mathematics minor, as well, so I 
can tell you you are wrong on two counts.  This test winner, test loser, 
test other loser terminology is confusing and unnecessary.  It is totally 
possible to write a computer program based on Mike's correct analysis 
that you only need to compare the pairwise results.  I have done it, and 
judging from Lucien's paper, Lucien did it as well.

I would ask that any examples from now on be examples that can be entered 
into mine or Lucien's or somebody's program for calculating Condorcet 
winners.  Feel free to write your own program that uses TW TL TOL 
notation, but please give us a tool to understand this.  Your examples 
are almost totally obfuscated by your notation, and many people I'm sure 
don't have the patience to wade through them, nor the tools to interpret 
them.

Thank you. 

Rob Lanphier
robla at eskimo.com
http://www.eskimo.com/~robla