On Mon, Dec 18, 2000 at 10:29:06AM -0600, Manoj Srivastava wrote: > Actually, since I had disconnected during the long debate that > progressed here, there was a lot of material to cover, and > digest. And I still think I would not be able to defend the method to > someone uninitiated; I need to go look for the papers on the voting > methods to be sure of my footing on this. > > However, I am willing to trust Raul and AJ to have researched > the underlying method, and the language is clean and understandable, > and thus I am willing to be a second for this proposal.
I'm still waiting for AJ to agree that the underlying method is good. As I understand it, he agrees with just about everything in the vote tallying method I proposed, except the handling of supermajority. Last I heard, AJ was thinking about proposing an amended version of my proposal such that when supermajority is relevant different voters having differing voting strengths. As far as theory goes: the Smith/Condorcet mechanism produces the same outcome as our current mechanism, except in its handling of certain kinds of ties. If there's a circular tie, it tosses out the weakest link of the chain, until the chain breaks. The current mechanism can be made to ignore circular ties -- essentially, you have to use A.6(2..4) only when there's a tie among first-preference options (and A.6(6) when there's an all-around tie). [Also, to properly handle votes with mixed supermajorities, you have to eliminate the "final" from A.6(7).] I couldn't think of a good way of phrasing A.6(2..4) so they'd only take effect when there's a first preference tie -- all my attempts took more verbiage than my original attempt at Smith/Condorcet. [Of course, I've fleshed out the Smith/Condorcet procedure since then...] If you want me to give this another shot, I can give it a go. Concorde (Single Transferrable Vote) biases the vote in the direction of what people say is best, using other preferences only to resolve ambiguity. Smith/Condorcet biases the vote a bit more in the direction of what the most people vote for. Personally, I don't really care which underlying mechanism is used: Smith/Condorcet or Concorde (Single Transferrable Vote). I just want something that's not ambiguous about vote counting. Thanks, -- Raul
