I realized after sending this message that I was unfair to winning-votes Condorcet methods like Schulze. After playing around with some scenarios, I now see that such methods do successfully elect A in spite of B's truncation. There are a few problems with this, though. The first is legitimacy: C voters would certainly squeal that they have more first preferences than A has votes. The second is, what if the truncation was honest; then, B would be the correct social utility winner. The third is, if instead of just truncating, the B voters actually bury A, then the dilemma returns.
So, Schulze is better than I thought; but I still think that runoff methods are the best solution. Perhaps a runoff between the Schulze-Margins and the Schulze-WV winner would be ideal - if that weren't hellishly impossible to explain. JQ
---- Election-Methods mailing list - see http://electorama.com/em for list info
