Very interesting, Jobst. Suppose that ratings are between zero and 100. On each ballot normalize these ratings by shifting to the left by its mean rating and then dividing by the root mean square of those ratings. This ballot contributes its normalized ratings to the respective candidates. [By convention if the standard deviation is zero, then it contributes the same value zero to each candidate.] It has been claimed (and seems plausible) that this version of CR satisfies your 0IMC criterion. Forest
________________________________
<<winmail.dat>>
---- Election-methods mailing list - see http://electorama.com/em for list info
