[EM] SCRRIRVE; my folly

James Green-Armytage jarmyta at antioch-college.edu
Tue Jan 20 13:43:02 PST 2004

Dear Chris Benham, et al, 
	Actually, it seems to me that SCRRIRVE might well be Condorcet efficient.
Which totally invalidates almost everything I said in my last post. (Or at
least invalidates it with respect to CCSCRRIRVE, although it is probably
still relevant with respect to other methods.)
	Here's why I think it's probably Condorcet efficient: Just as an IRV
winner must have a pairwise beat over at least one other candidate, I
think that a reverse IRV winner (an IRV winner with reverse rankings)
might have to have at least one pairwise loss against another candidate.
Thus SCRRIRVE can't eliminate a Condorcet winner. Anyway, if it's not
Condorcet efficient, then it's dern close. I don't know what the effect of
the symmetrically created ballots are as far as Condorcet efficiency.
Also, I don't know whether it's efficient with respect to Schwartz, Smith,
or Condorcet loser.
	Anyway, if SCRRIRVE is Condorcet efficient, then there is no need to call
	And if SCRRIRVE is Condorcet efficient, then it doesn't fall into the
"hybrid" category I described, but rather into the other category, of
"pure" Condorcet methods or whatever. In that case, strategies to steal
the election from a sincere Condorcet winner need to revolve around
creating a false pairwise beat which overrules a sincere pairwise beat.
	Burying is still a significant problem in SCRRIRVE, but as for whether it
is more or less of a problem than in WV, I'll have to go into that later,
since I don't have much time just at the moment. I note in passing that
SCRRIRVE seems to have some things in common with Raynaud (eliminate the
candidate who has the strongest beat against them and recalculate, until
there is an unbeaten candidate), since they are both elimination-based
Condorcet methods. Perhaps someone else (Markus?) can tell us why Raynaud
isn't discussed more often. I imagine that there is a good reason for it,
since Raynaud is just about as obvious as WV minimax / sequential
dropping, and thus must have been discussed at some point.


More information about the Election-Methods mailing list