[EM] Dyadic approval implemented as CR

Roy royone at yahoo.com
Thu Aug 23 08:36:46 PDT 2001

CR is a great system in theory, but strategic voting makes it 
degenerate into Approval. Dyadic Approval provides multiple levels of 
approval, but the ballot is complicated. Can they be synthesized?

How about a CR ballot that is scored as Dyadic approval? Votes in the 
top half of the ratings scale score 1 point (in a pairwise matrix) 
over those in the bottom half. Within a half, those in the upper half-
of-the-half score a half point over those in the lower. And so forth. 
Sum the matrices, run a Condorcet method.

An alternative scoring method could be to take the difference between 
the candidates' ratings, as a proportion of the whole range, and 
round up to a power-of-1/2. If the scale goes from 1 to 100, a 
difference of 50 or more scores 1, a difference of 25-49 scores 1/2, 
etc. That way, a score of 48 compared to a 51 isn't as great as a 
score of 90 vs. 10, but you still have the coarse-grained counting 
that doesn't make everything gravitate to the extremes.

First choice will be 100, last choice will be 1, and others will be 
in-between, because your first choice scores just as much against 
unfavored choices no matter where they're ranked in the lower half.

One other alternative scoring method would be to run it as Approval 
Run-Off, although that might be cumbersome. I haven't given it much 
consideration yet.

Roy Johnson

More information about the Election-Methods mailing list