Two apples-Chocolate

[Steve Eppley]

Demorep1 wrote:
[snip]
>With the plain Condorcet tie breaker (based on which worst defeat
>has the lowest margin) C wins. 
[snip]

It appears you haven't been paying close attention to the discussion
of the definition of Condorcet.  You may want to go back and read
those messages (from about 3 or 4 weeks ago) if you missed them.  

I was also under the impression that size of pairwise defeat meant
the margin of defeat, but this isn't true.  Condorcet measures the
defeat of the pair-loser by the number of votes for the pair-winner.

Suppose a 3-candidate election has the following circular tie:
  B loses to A   (40 to 50)
  C loses to B   (49 to 51)
  A loses to C   (45 to 49)

The real definition of Condorcet ignores the margin, and says that 
the sizes of the pairwise defeats are:
  B by 50
  C by 51
  A by 49.
So A is the tie-break winner by Condorcet, even though C's margin of
defeat in C's loss is smaller than A's margin of defeat in A's loss.

You may not like this definition of Condorcet, thinking that margin
is a fairer way to measure size of defeat.  But that doesn't mean 
it's proper to use the name Condorcet for the margin method.

There was also some brief discussion in EM a couple of weeks ago
about why the margin method is inferior to Condorcet.  The reason
was that it gives supporters of one candidate (like Dole in Mike O's
example which demonstrated offensive order-reversal) the tactical
option of swinging the result in that candidate's favor by manipu-
lating margins, or manipulating to create a circular tie where that
candidate has the smallest margin.

I'm privately forwarding a message which shows an example of this
possible manipulation to Demorep1.  It's one of Mike O's
Dole/Clinton/Nader scenarios.

--Steve