[EM] First Condorcet cycle ever spotted in a national presidential election (!?! apparently)

[Kristofer Munsterhjelm]

Warren Smith wrote:
> I am currently working on writing  a paper on this election.  In the
> event you or anybody can find more poll data, that'd help.  I'd be
> especially interested if anybody can find a range-voting-style poll.
> I'm handicapped by lack of knowledge of the Romanian language.

If I'm doing this right... using your poll data and the final election 
result, one gets:

BAS vs ANT: 47-53 (Insomar poll)
GEN vs ANT: 52.4-47.6 (ditto)
BAS vs GEN: 50.33-49.66 (actual election result)

which gives a Condorcet matrix of (where row beats column)

     #ANT #BAS #GEN
ANT  0.0 53.0 47.6
BAS 47.0  0.0 50.3
GEN 52.4 49.7  0.0

where the advanced methods thus give GEN > ANT > BAS. In other words, 
the candidate who won the actual election would have been ranked last 
under these methods!