[EM] “¡One can introduce advanced voting systems to ponies, but one cannot make the ponies implement the advanced voting systems!”

⸘Ŭalabio‽ Walabio at MacOSX.Com
Wed Dec 5 22:46:31 PST 2012


	2012-12-05T16:40:13Z, Kristofer Munsterhjelm:

> 	On 12/03/2012 05:53 AM, ⸘Ŭalabio‽ wrote:

> 	I only read the beginning of that thread, but it seems they're agreeing with RBJ: Score/Range asks too much of the voter.

	My preferred method is Score-Voting, but I suggested ranked ballots with 6 ranks because we have 6 mares running for Best Mare as an acceptable alternative.  This is the history:

	We initially used plurality, but it did not work.  The solution is to use truncated Borda-Count with 2 ranking with # 1 getting 2 votes and # 2 getting 1 vote.  I suppose that if this does not work, we can try truncated Borda-Count with 3 ranks.  This crazy incrementalism frustrates me.

	If we would use full rankings, we could use Condorcet, Oklahoman Electoral Primary-System, and Borda-Count on the output and get 3 full rankings as a result.  That would be interesting

> 	In any event, their objections (or arguments) seem to be based on the difficulty of answering honestly, not on strategic concerns.

	Yes, these are honest ponies.  I suspect that because of their honesty, Borda-Count and Condorcet would both elect the same Best Mare.

> 	I'd say there's another method that fits in between and isn't too complex: MJ. MJ doesn't use ratings (and clearly not ratings from -999...999) but grades like "Poor", "Mediocre", "Good", etc.

	The thing about Majority-Judgement is that the more detailed it is, the less accurate it is.  Still, what Majority-Judgement does is the equivalent of  filtering out outliners which might change the result:

	We all know the 80%/20%-rule.  This relates to the 1st Feigenbaum-Constant (4.66920160910299067185320382046620161725818557747576863274565134300413433021131473713868974402394801381716).  Let us suppose that we rank the mares on scale from negative -99 to positive +99, then delete all scores of negative -99 and positive +99, and then remove the top 20% of scores for each mare and the bottom 20% of scores for each mare, and then average.  The result should be mathematically equivalent to Majority-Judgement.

	1 advantage I see to this, is that it should favor centrists as much as Borda-Count, but be as resistant to tactical voting as Condorcet and Approval Voting.  It is an interesting proposal.

> 	And if grading is too hard, then Schulze should work. It's hard to count, but there are websites that will do the actual counting.

	I suggested doing full rankings and doing Borda-Count, Condorcet, and Oklahoman Electoral Primary-System on the results.


More information about the Election-Methods mailing list