>M.Rouse: >Very interesting method. Just a quick question. The article mentions that >rank-order methods would be really tough to do. Wouldn't the method just >be two votes with the correct rank-order and one with the reverse order?
--that is an interesting point. I had been imagining handling Condorcet or Borda by voting on every candidate-pair relationship (which would be a nightmare, especially with 50 candidates, and considering the task of ballot validity checking). Rouse's point is that we can just reverse EVERY pair at once en masse, by supplying a reverse-order ballot. Thus enabling Borda and Condorcet still to be handled apparently fairly reasonably (although IRV seems still not to be amenable). But before patting yourself on the back, consider these objections. 1. the Rouse reversed ballot might really make no political sense even though the forward ballot makes sense. If so, then it would be easy for the government to tell which one was the reversed one. That might enable them to cheat with more success probability, although it'd still be extremely risky for the govt to cheat, so I'm not too worried except in small elections. 2. In a 50-candidate election (where the Rouse-reversal plan pays off big versus the stupid all-pairs plan) there are 50! possible votes and that far exceeds the number of voters (or for that matter the number of atoms). Therefore, it will be easy to MATCH UP the forward & backward ballots, and many people's votes will be UNIQUE. I.e. the whole idea of anonymity will be lost. If you come forward with your vote-copy receipt, it will be obvious to all what your vote-triple was. Secrecy of the ballot is lost. I think criticism number 2 is a very serious criticism and pretty much blows your plan out of the water. However... we could try to reintroduce it to the water by having a machine chop up every A>B>C>D ballot into A>B, B>C, and C>D ballots? No good, that omits A>C and A>D and B>D implications. So... looks grim, sorry. wds ---- election-methods mailing list - see http://electorama.com/em for list info
